Towards Near-Real-Time Damage Estimation Using a CSM-Based Tool

Journal of Earthquake Engineering, 12(S2):199–210, 2008

Copyright � A.S. Elnashai & N.N. Ambraseys

ISSN: 1363-2469 print / 1559-808X online

DOI: 10.1080/13632460802014055

Towards Near-Real-Time Damage Estimation Usinga CSM-Based Tool for Seismic Risk Assessment

DOMINIK H. LANG1, SERGIO MOLINA-PALACIOS2,and CONRAD D. LINDHOLM1

1NORSAR, International Centre for Geohazards (ICG), Kjeller, Norway2Departamento de Ciencias del Mar y Biologıa Aplicada, Universidad

de Alicante, Alicante, Spain

Loss estimation studies have traditionally been using individual scenario earthquakes as the basisfor planning and decision making. Existing knowledge of regional geology and seismology have, inthe same context, been used to generate maps with estimated intensities or ground-motion para-meters (e.g., accelerations). In combination with other factors (e.g., composition and structuralcharacteristics of the building stock), these parameters allow a direct estimation of the damage tostructures and lifelines and of the impacts on the population (e.g., number of injured and fatalities).Nowadays, the advent of high-speed computing, satellite telemetry, and Geographic InformationSystems (GIS) has made it possible to generate rapid loss estimates for multiple earthquakescenarios, and to provide a greatly improved mapping capability. Perhaps most importantly, it isnow within reach to derive damage and loss estimates for an earthquake in near real-time solelybased on the source parameters of an event, with magnitude and location as a minimum. We haveinvestigated the capabilities and efficiency of the seismic risk and loss assessment tool SELENA[Molina et al., 2007] for such a near real-time application. One of the main features of this toolconsists of taking into account epistemic uncertainties in all types of input data by a logic-treecomputation scheme. Therefore, the influence of the number of logic-tree branches on the computa-tion time is examined in order to evaluate its applicability for near-real-time damage assessment.

The SELENA software tool is tested in a near-real-time context for the Arenella district inNaples (Italy), one of the test cities within the EC-SAFER project. The risk and loss assessmentstudies depend on real-time shake map estimates along with the building stock inventory facilitatedby the University of Naples as well as structural capacity curves and vulnerability functionsdeveloped for the predominant building types by Polese et al. [2008].

Keywords Seismic Risk and Loss Assessment; Near-Real-Time Damage Estimation; Capacity-Spectrum Method; Scenario Earthquake; Naples

1. Introduction

The developments within earthquake engineering over the past decade (e.g., ATC, 1996;

Freeman, 1998; Fajfar, 1999) have significantly improved the capacity to analytically and

empirically evaluate building performance during earthquake shaking. This has initiated

a number of earthquake damage scenario studies [Ordaz et al., 2000; Bommer et al.,

2002; Carvalho et al., 2002, Erdik et al., 2003; Molina and Lindholm, 2005; and others)

for different levels of resolution, in which the seismic risk was derived for single

buildings, certain building typologies, city districts, or entire communities. It is possibly

because damage scenario results can be presented in intuitive figures and colored maps

Address correspondence to Dominik H. Lang, NORSAR, International Centre for Geohazards (ICG),

P.O. Box 53, Instituttveien 25, 2027 Kjeller, Norway; E-mail: [email protected]

199

that these developments have provided the scientists with tools which can be used in city

planning contexts as well as for consciousness mobilization [Eguchi et al., 1997; FEMA,

1999, 2004; Molina and Lindholm, 2006]. The easy-to-grasp damage distribution maps

are well suited to convey the basic message of earthquake risk and loss studies, and for

this reason many hazard study results are exclusively presented in the form of maps. One

of the great advantages of such maps is that the results presented in this way may be

understandable by many while still maintaining professional quality.

Recently, these tools have been applied in a new setting. Through the rapid (auto-

matic and semi-automatic) determination of hypocenters and even fault ruptures follow-

ing a large earthquake, near-real-time shake maps illustrating the areal distribution of

shaking effects (i.e., intensity, ground acceleration) can be generated and provided to

emergency and civil protection agencies as well as to the general public through web

portals (e.g. USGS ShakeMaps; Wald et al., 1999). A further step in the same direction is

to compute (at different levels of resolution) damage and death-toll estimates (e.g. Wyss,

2005), a type of information which is aimed at supporting the rescue work.

Among the objectives for near-real-time damage scenarios, one of the most impor-

tant is that after a disastrous earthquake even a reasonable damage information overview

is difficult to obtain during the first hours or even days, such as in the case of the October

2005 Kashmir earthquake. During the first days and weeks following the earthquake,

overview maps that indicate relative and absolute damage distribution may be of great

importance for rescuing lives and property, and for providing relief [Wyss, 2006].

2. Methods and Approaches for Seismic Risk Estimation

Two projects that were running more or less in parallel in the 1990’s demonstrate how

two different approaches were chosen:

� An empirical approach was used in the RADIUS project which focused on cities at

risk (http://www.geohaz.org/contents/projects/radius.html ). For the selected cities,

scenario earthquakes and ground shaking in terms of intensity using the Modified

Mercalli Intensity Scale (MMI) were defined and Damage Probability Matrices

(DPM) were developed.

� An analytical approach was used in the HAZUS project [FEMA, 1999, 2004],

which consumed considerably more resources and which focused more on an

engineering-wise characterization of a wide range of model building types. For

147 different building types (classified in terms of building materials, construction

type, building regulatory regime, etc.), building capacity curves and respective

fragility functions were derived so that damage estimates could be calculated

quantitatively for predicted ground-motion estimates.

The method applied in the following is concentrated on the latter approach which is a

modification and further development of the HAZUS-MH methodology [FEMA 1999,

2004] based on the capacity spectrum method (CSM). Using the core of this method

within a logic-tree computation scheme in order to account for epistemic uncertainties, an

open-source tool was developed under MATLAB� and called SELENA (SEimic Loss

EstimatioN using a logic tree Approach; Molina and Lindholm, 2006). Currently,

SELENA computes median values as well as 16 and 84% fractiles of the risk results.

This is done by means of a logic-tree methodology in which the different branches of the

tree can be weighted so that at the end of the computation, the risk results are multiplied

by their corresponding weights and fitted to a normal distribution in order to get the

median value as well as its range of uncertainty (standard deviation). The single branches

200 D. H. Lang, S. Molina-Palacios, and C. D. Lindholm

of the logic tree may represent uncertainties in the source parameters of a deterministic

seismic event or the considered probabilistic shake map, the applied ground-motion

prediction relationship, soil type, building capacity curves and fragility functions, and

the economic loss model.

Running SELENA under MATLAB� provides the user with some advantages

such as full flexibility with respect to input and presentation of results as well as

full transparency of the code in order to make own modifications and amendments.

Moreover, SELENA facilitates the estimation of the structural damage to the build-

ing stock, direct economic losses as well as human losses (i.e., casualties) related to these

physical damages. All results are provided with uncertainties on the census tract level,

which is the minimum geographical unit to be defined by the user (e.g., city districts,

quarters) and for which a number of basis information is required (e.g., soil condition,

building stock composition, etc.).

3. Near-Real-Time Risk Assessment Based on Rapid Event Detection

Figure 1 schematically illustrates the processing scheme for a near-real-time risk and loss

assessment as it is applied herein. Even though the basic underlying approach of our

procedure is totally different from the EPEDAT tool [Eguchi et al., 1997], the operation

sequences of both show some similarities.

Following an earthquake, the seismic network will be the first entity detecting it and

sending the information to the different national or international seismological observa-

tories where a rapid automatic location process is initiated. Irrespective of the fact that in

most cases a more precise determination of the source parameters is conducted with some

time delay, the parameters of the first event detection may be used for the near-real-time

process. In case of a larger magnitude event, it may be necessary to identify the extent of

the fault rupture in addition since ground-motion characteristics will strongly depend on

the rupture length and the distance between the fault rupture plane and the (damaged)

area of interest. This is naturally more critical for those events that comprise multiple

segment ruptures [Eguchi et al., 1997].

In general, it should be noted that the ground shaking estimation process for an

estimated region will be different mainly depending on three aspects:

� level of seismicity (high-, intermediate-, and low-seismicity regions);

� population density (densely populated, sparsely populated, or remote regions);

� development status and level of prosperity.

The level of seismicity mainly determines the question of whether or not experi-

ence with local earthquake disasters and knowledge on the seismic hazard is already

available in a given region. In case of high-seismicity regions, for example, the

location of the causative fault (source) may be mapped in detail and its seismic hazard

may be so well known that a kind of ‘‘scenario library’’ can be produced a priori,

containing the ground-motion shake maps for the respective city or municipality at risk

(Fig. 1). This scenario library could contain not only shake maps of recent earthquakes

but also for events in the past processed in order to allow a comparison with recent

events [Wald et al., 1999]. In addition, information on damage statistics during past

events could be used to calibrate the assumptions for the risk scenario such as

geological soil conditions or applied ground-motion prediction relationships. In con-

trast to that, a scenario library will be difficult to establish for areas with low or

intermediate seismicity, where much less is known about the earthquake scenarios that

are likely to occur.

Towards Near-Real-Time Damage Estimation Using a CSM-Based Tool 201

Along with the level of seismicity, population density and the development status of

a given region mainly determine the availability and quantity of seismic recording

(strong-motion) stations. It should be stressed that realistic input data of locally recorded

earthquakes will increase the reliability of the study and furthermore allow for a calibra-

tion of the assumptions made for the deterministic earthquake scenario. As illustrated by

the flowchart in Fig. 1, it has to be checked whether recording stations are available

within or close to the study area. If so, the recorded instrumental ground-motion data may

be used to complement the scenario library and to adjust the derived deterministic shake

maps. It is likely that the locations of possibly available recording stations will not

comply with the specified geographical units (i.e., census tracts) which represent the

basis for the risk assessment. Consequently, the provided spectral ground-motion ordi-

nates have to be assigned to the centers of these units (centroids). The procedure applied

for this, as presently incorporated in SELENA, examines which stations are within a 5 km

radius around each centroid (see Fig. 2). If at least 5 points meet this criterion, the mean

value and corresponding 16%, respectively, 84% fractiles of the spectral ordinates are

SEISMOLOGICAL OBSERVATORY:

location, depth, magnitude

no

yes

no

computation of deterministic shake map

output: ascii-files facilitated for mapping

SELENA

Stations closeto study area?

yes scenario library

Included in scenario library?

SEISMIC NETWORK:

event detection

assign ground-motion to thecensus tracts (PGA, Sa)

provision of deterministic shake map

real-time shake map basedon instrumental data

computation ofeconomic losses

computation ofhuman losses (casualties)

computation ofbuilding damage

FIGURE 1 Flowchart of a near-real-time damage and loss assessment procedure using

the software tool SELENA. The components in the shaded box indicate the computation

steps to be conducted within SELENA.


computed and assigned to the respective centroid. If less than 5 points are within this

5 km radius, a new circle of 10 km diameter is drawn.

In cases when neither strong-motion stations exist in the area nor a scenario library

could be prepared due to the low seismicity of the study area, the determination of

seismic ground-motion parameters can solely be realized by the computation of shake

maps based on a deterministic scenario using the provided earthquake source information

(Fig. 1). Once the shake maps are generated or compiled, the basic computation part

within SELENA starts and the direct structural damages, economic, and human losses are

subsequently computed. It should be recognized that the computation process as illu-

strated in Fig. 1 requires a considerable amount of static input data on the study area’s

soil conditions, building stock, and population. All input data is to be prepared and

provided in terms of ascii-files. Since these data in principal are static information it

can be prepared in advance and possibly updated regularly.

A detailed description of SELENA’s background and computational principles is

given in Molina et al. [2007]. SELENA is based on the capacity-spectrum method [ATC,

1996; FEMA, 2005], which is an iterative procedure in order to find the performance

point, i.e., the spectral displacement a structure undergoes during lateral earthquake

impact, accounting for its nonlinear behavior. It should be noted that SELENA’s compu-

tational process both due to the iterative calculations and due to the logic-tree scheme

takes longer than a ‘‘conventional’’ calculation. Since the computation of the perfor-

mance point depends on the seismic demand (represented by the design spectrum in

SASD-format) and the building’s nonlinear behavior under lateral loading (represented

by the structural capacity curve), this iterative computation is performed separately for

each census tract and each building type.

As a consequence of this, the question of whether or not the whole procedure is able

to meet the requirements of a real-time or near-real-time damage assessment is basically

dependent on the following two aspects:

FIGURE 2 Assignment of strong-motion stations to the centers of the geographical units

(centroids) in order to provide the spectral ground-motion ordinates.


� the readiness/rapidness of the event detection and its magnitude and location

determination, and

� the software’s actual computing time which is mainly determined by the size of

the study area (i.e., the number of census tracts, number of buildings, and/or

variety/diversity of building types) and the defined branches of the logic tree in

order to account for uncertainties in input parameters.

Needless to say, an increasing number of the logic-tree branches increases the

computation time and thus the availability of results significantly, and hence it is

necessary to check the impact of the level of refinement (i.e., the number of logic tree

branches) on the analysis results (i.e., predicted damages and losses). In the following,

this will be conducted for the Arenella district in the metropolitan area of Naples, Italy.

4. Screening Tests of the Procedure Towards a Near-Real-Time Application

4.1. Test Study Area: Arenella District of Naples, Italy

Due to its location close to the active part of the Campanian volcanic arc, the city of

Naples is exposed to one of the most significant volcanic hazards in Southern Europe.

Values for Peak Ground Acceleration of 0.15g with a 10% probability of exceedance in

50 years are expected for the city [Giardini et al., 2003]. Even though the level of

seismic hazard is only moderate, the combination with its housing density and the fact

that the majority of buildings are older than the seismic code provisions, Naples can be

regarded as one of the high-risk areas in Italy. Moreover, the Naples metropolitan area,

with approximately 4.4 million people, is the second most populated in Italy. The city

itself has a population of around 1 million and covers an area of 117 km2. It is divided

into 10 municipalities (municipalita) which in turn are subdivided into 30 districts

(quartieri).

The district of Arenella which was chosen for the present study has an area of 5.25

km2 and 72,000 inhabitants. As Fig. 3 illustrates, the Arenella district is subdivided into

129 different census tracts (i.e., minimum geographical units) for which the seismic risk

assessment is carried out. Compared to other parts of the city, the Arenella district is

distinguished for the study by its high quality of input information in terms of population

statistics and building stock inventory (University of Naples Federico II).

The building stock of the Arenella district consists of approximately 1,500 individual

buildings which show only slight variations in terms of the buildings’ years of construc-

tion, occupancy, as well as structural systems (type of materials). More than 55% of the

buildings were built between 1945 and 1970, while around 80% are of residential use and

nearly 70% have a story number � 4. Since more than 87% of the building area in the

Arenella district contain reinforced-concrete frame structures [Polese et al., 2008], the

risk assessment is only concentrated on these building types. A sub-classification of these

buildings is done according to the number of stories. Consequently, we are distinguishing

between RC frame structures with 1–3 (RC 1), 4–6 (RC 4), and 7 or more stories (RC 7).

For these classes, Polese et al. [2008] developed structural capacity curves and vulner-

ability functions which establish the basis for the risk and loss assessment.

The geological soil conditions are assumed to be homogeneous over all census tracts

and range between NEHRP-site classes C (average shear-wave velocity vs,30 = 360–760 m/s)

and D (vs,30 = 180–360 m/s). In order to account for local soil amplification effects, both site

classes will be considered during the risk computation procedure by a corresponding logic

tree formulation.


4.2. Risk and Loss Assessment for the 1980 Irpinia Earthquake Scenario

Since the Irpinia earthquake (Ms 6.9) of November 23, 1980, was the most recent

damaging event in the Campania region, its source parameters serve as the main basis

for the deterministic earthquake scenarios. The epicenter of the 1980 earthquake was

located approximately 95 km east of Naples and was therefore not able to produce

any significant damages to the building stock in the Arenella district. In this context,

the epicentral coordinates are relocated and shifted closer to the study area. Thereby

point sources are assumed such that the influences of directivity or fault length

effects are not covered. Risk and loss results may be altered if these effects would

be incorporated in defining the hazard. To compute the spectral ground-motion

parameters for the Arenella district, an empirical ground-motion prediction relation

is applied which was developed based on Italian strong-motion data [Sabetta and

Pugliese, 1996].

Figure 4 graphically illustrates the extent of structural damages to the three

different RC building types as well as the total economic losses for the four different

earthquake scenarios. This was done by relocating the epicenter of the 1980 earth-

quake from its origin in the Campano-Lucano Apennines at distances of 15, 25, 40,

and 60 km from the Arenella district. From Fig. 4 it is obvious that mid- and high-

rise building types RC4 and RC7 show much larger damages than the low-rise RC1

buildings and for all three building types the decrease in damage with increasing

epicentral distance Re is well illustrated. The differences in the predicted economic

losses between these four scenarios show that the loss is nearly doubled as distance

Re is reduced by 10, 15, or 20 km.

FIGURE 3 Administrative division of the city of Naples into 10 municipalities (muni-

cipalita) and 30 districts (quartieri; partly shown); the Arenella district is further sub-

divided into 129 census tracts.


With respect to a near-real-time application of SELENA, the total computation time

on a conventional PC was 58 s for the 129 census tracts (and 3 building types). For this

scenario, the logic tree only consisted of two branches.

4.3. Influence of Branching on Computation Time and Stability of Results

Based on the above deterministic scenario of the (modified) 1980 Irpinia earthquake with

its epicenter fixed to an epicentral distance Re = 25 km from the Arenella district, the

impact of the number of logic-tree branches on the loss results, their stability, and the

computation time is investigated.

Figure 5 illustrates the differences in predicted losses between four different risk

computations which rely on different representations of the scenario earthquake’s mag-

nitude. It is obvious that the differences are large, especially when comparing the losses

FIGURE 4 Percentages of predicted building damage and total estimated losses for the

Arenella district dependent on epicentral distance Re.


for the unmodified case (eq1) and the cases accounting for uncertainties in the earthquake

magnitude (eq3, eq5, and eq9). In contrast, differences in losses between eq3 and eq5,

between eq3 and eq9, and between eq5 and eq9 are small. It can be concluded that the

quality of results significantly improves with only 3 branches (eq3). However, adding

more branches does not lead to a refinement of results while computation time is

extended dramatically. This, however, justifying limitations in the defined number of

logic-tree branches with respect to the definition of the scenario earthquake.

Besides the above, it should be noted that for each of the four comparisons, the

differences in losses between the 129 census tracts are similar and characterized by only

small variations. This feature reflects the nearly-equal distances of all census tracts to the

source and the homogeneous conditions in all census tracts (in terms of subsoil conditions

and building stock).

A similar effect as shown in Fig. 5 can be observed when branching is done

simultaneously over more types of input data such as applied ground-motion prediction

equation or soil classes. The differences between the median-loss results diminish as the

number of branches increases, which in turn justifies a limitation of the logic tree

branches to a reasonable number in order to keep the computation time short.

The impact of the number of branches on the computation time is demonstrated in

Fig. 6 for the Arenella district with 129 census tracts and approximately 1,500 individual

buildings. Since both the computation time and the number of branches are linearly

connected it is certainly desirable to expel as many uncertainties as possible in the

forefront in order to reduce the computation time at the end.

5. Conclusions

In order to comply with the needs of rapid information and to conduct a near-real-time

damage estimation, the computation time itself must be restricted. This can only be done

FIGURE 5 Differences in predicted losses between different ways of defining the deter-

ministic scenario earthquakes: eq1 = one earthquake (M 6.9) weighted (1.0), eq3 = three

earthquakes (M 6.5, 6.9, 7.3) weighted (0.3, 0.4, 0.3), eq5 = five earthquakes (M 6.5, 6.7,

6.9, 7.1, 7.3) weighted (0.15, 0.2, 0.3, 0.2, 0.15), eq9 = nine earthquakes (M 6.5, 6.6, 6.7,

6.8, 6.9, 7.0, 7.1, 7.2, 7.3) weighted (0.05, 0.1, 0.1, 0.15, 0.2, 0.15, 0.1, 0.1, 0.05).


by limiting the number of branches of the logic tree computation scheme. Since most of

the required input is static such information can be prepared and updated at regular

intervals before a damaging event occurs. This concerns, in particular, the input data

connected to soil conditions, building stock inventory, structural capacity curves and

fragility functions, population statistics, and the economic loss model. With respect to the

latter, it should be noted that this type of input information is of minor interest in a real-

time application where priorities are more concentrated on damage assessment and

casualty estimation in order to prepare the affected society for the aftermath and to

keep secondary losses low. Obviously, the most uncertain input parameters are those

related to the earthquake source (i.e., magnitude and location) as well as the applied

ground-motion prediction model. However, since these attenuation relations are generally

constricted to a certain magnitude-distance range and/or fault mechanism, they can be

pre-selected according to the expected earthquake parameters and made available for the

study. The predicted structural damages and losses for the Arenella district (Fig. 4) which

were computed for the 1980 Irpinia earthquake demonstrate how earthquake source

parameters (here epicentral distance) strongly influence the loss results. Consequently,

both the time span needed to detect the source parameters of an earthquake and the

reliability of these estimated parameters are shown to be crucial in a near-real-time

damage estimation. The conducted tests on different levels of uncertainty showed that

lower numbers of logic-tree branches do provide reliable results while reducing the

computation time. This should of course be balanced against the fact that more branches

of the logic tree will provide more sound confidence intervals for the final damage

estimates.

Acknowledgments

The present risk study for the Arenella district is based on a thorough building stock

inventory and comprehensive structural analyses which were conducted by the University

FIGURE 6 Computation time as a function of numbers of branches in the logic tree.


of Naples Federico II. We are very grateful to Maria Polese, Iunio Iervolino, and Gaetano

Manfredi for their efforts and for providing us with these data. Thanks also to Hilmar

Bungum and two anonymous reviewers for their comments, which led to significant

improvements to the article. We also acknowledge the support from the EC-SAFER

project (STREP no. 036935) as well as VIGROB-009 and GV07/045. The development

of the SELENA tool is supported by the International Centre for Geohazards (ICG).

References

ATC [1996] ‘‘ATC-40 – Seismic Evaluation and Retrofit of Concrete Buildings,’’ Vol. 1, Report

No. SSC 96–01, Applied Technology Council, Redwood City, CA.

Bommer, J., Spence, R., Erdik, M., Tabuchi, S., Aydinoglu, N., Booth, E., del Re, D., and Peterken, O.

[2002] ‘‘Development of an earthquake loss model for Turkish catastrophe insurance,’’ Journal of

Seismology 6, 431–446.

Carvalho, E. C., Coelho, E., Campos-Costa, A., Sousa, M. L., and Candeias, P. [2002]

‘‘Vulnerability evaluation of residential buildings in Portugal,’’ Proceedings of 12th European

Conference on Earthquake Engineering, London.

Eguchi, R. T., Goltz, J. D., Seligson, H. A., Flores, P. J., Heaton, T. H., and Bortugno, E. [1997]

‘‘Real-time loss estimation as an emergency response decision support system: The early post-

earthquake damage assessment tool (EPEDAT),’’ Earthquake Spectra 13, 815–832.

Erdik, M., Durukal, E., Siyahi, B., Fahjna, Y., Sesetyan, K., Demirciglu, M., and Akman, H. [2003]

‘‘Earthquake risk mitigation in Istanbul,’’ In: Mulargia, F. and Geller, R. J. eds. Earthquake

Science and Seismic Risk Reduction, NATO Science Series IV. Earth and Environmental

Sciences, Vol. 32 (Kluwer Academic Publishers, Dordrecht), pp. 250–283.

Fajfar, P. [1999] ‘‘Capacity spectrum method based on inelastic demand spectra,’’ Earthquake

Engineering and Structural Dynamics 28, 979–993.

FEMA [1999] ‘‘HAZUS 99 – Earthquake Loss Estimation Methodology. Technical Manual,’’

Federal Emergency Management Agency, Washington D.C.

FEMA [2004] ‘‘HAZUS-MH – multi-hazard loss estimation methodology, earthquake model,

advance engineering building module,’’ Federal Emergency Management Agency,

Washington D.C.

FEMA [2005] ‘‘FEMA 440 – improvement of nonlinear static seismic analysis procedures,’’

Federal Emergency Management Agency, Washington D.C.

Freeman, S. A. [1998] ‘‘Development and use of capacity spectrum method,’’ Proceedings of the 6th

U.S. National Conference on Earthquake Engineering, Seattle, Washington.

Giardini, D., Grunthal, G., Shedlock, K. M., and Zhang, P. [2003] ‘‘The GSHAP Global Seismic

Hazard Map,’’ In: Lee, W., Kanamori, H., Jennings, P., and Kisslinger, C. eds. International

Handbook of Earthquake & Engineering Seismology, International Geophysics Series 81 B,

Amsterdam: Academic Press, 1233–1239.

Molina, S. and Lindholm C. D. [2005] ‘‘A logic tree extension of the capacity spectrum method

developed to estimate seismic risk in Oslo, Norway,’’ Journal of Earthquake Engineering 9(6),

877–897.

Molina, S. and Lindholm, C. D. [2006] ‘‘A capacity spectrum method based tool developed to

properly include the uncertainties in the seismic risk assessment, under a logic tree scheme,’’

ECI Geohazards – Technical, Economical and Social Risk Evaluation, Lillehammer, Norway.

Molina, S., Lang, D. H., and Lindholm, C. D. [2007] ‘‘SELENA v2.0 – User and Technical Manual

v2.0,’’ NORSAR, Kjeller, Norway.

Ordaz, M., Miranda, E., Reinoso, E., and Perez-Rocha, L. E. [2000] ‘‘Seismic loss estimation model

for Mexico city,’’ Proceedings of the 12th World Conference on Earthquake Engineering,

Auckland, New Zealand, Paper No. 1902.

Polese, M., Verderame, G. M., Mariniello, C., Iervolino, I., and Manfredi, G. [2007] ‘‘Vulnerability

curves for gravity load designed RC buildings in Naples – Italy,’’ Journal of Earthquake

Engineering, 12(S2), 234–245.


Sabetta, F. and Pugliese, A. [1996] ‘‘Estimation of response spectra and simulation of nonstationary

earthquake ground motions,’’ Bulletin of the Seismological Society of America 86(2), 337–352.

Wald, D. W, Quitoriano, V., Heaton, T. H., Kanamori, H., Scrivner, C. W., and Worden, C. B.

[1999] ‘‘TriNet ‘‘ShakeMaps’’: Rapid Generation of Peak Ground Motion and Intensity Maps

for Earthquakes in Southern California,’’ Earthquake Spectra 15(3), 537–555.

Wyss, M. [2005] ‘‘Earthquake loss estimates applied in real time and to megacity risk assessment,’’

Proceedings of the Second International ISCRAM Conference, Brussels, Belgium, Carle, B. and

van de Walle, B., eds. pp. 297–299.

Wyss, M. [2006] ‘‘The Kashmir M7.6 shock of 8 October 2005 calibrates estimates of losses in

future Himalayan earthquakes,’’ Proceedings of the Third International ISCRAM Conference,

Newark, Van de Walle B. and Turoff M., eds., pp. 5.


Documents

Towards Near-Real-Time Damage Estimation Using a CSM-Based Tool