Earthquake hazard, risk and loss
The project is engaged in research topics within the following fields:
- Quantification of earthquake hazard (exceedance probability of a certain ground shaking amplitude/level) including the influence of soil amplification characteristics
- Seismic vulnerability of the built and populated environment
- Development of models and software tools for risk quantification (risk = vulnerability * hazard)
- Loss modeling at different scales
Here, the term earthquake hazard refers to the occurrence probabilities of damaging ground motions, exclusively relating to natural phenomena and processes, while risk and loss results from combining the earthquake hazard with the vulnerability of the building stock.
Figure 1. Collapse of a dwelling due to the Kashmir earthquake of October 8, 2005.
Earthquakes hit without warning (Fig. 1), but with a damage potential that is generally confined to a limited area around the epicentre or along the fault trace. In most cases, it is the combination of earthquake magnitude, poor building quality, and high population density of the area of highest shaking that cause the disasters. This is one of the reasons why, with increasing population and urbanization in earthquake-prone countries, the losses from earthquakes have been increasing significantly over the last decades (Fig. 2). When comparing earthquake risk with other natural risks it is informative to see from Fig. 2 that, at low probabilities, the earthquake risk is far above the risk from other natural hazards (even in low-seismicity regions like e.g., Germany).
This implies that the earthquake damage increases strongly with decreasing occurrence probabilities (increasing return periods), which in turn means that the largest ones are rare but very destructive. This indicates that at any given location one cannot rely on `human lifetime memory¿ as a basis for precautionary measures: science is needed instead.
At present, earthquakes cannot be predicted within reasonable time and spatial windows, and the viable disaster prevention is therefore to delineate the earthquake-prone areas and to understand in detail the factors that turn an earthquake into a disaster or prevents the same.
Only with such knowledge, society can develop the capacity to limit future tragedies, and mitigation is a key word for the work within the ICG Project 3.
Figure 2. (left) Global economic losses per decade in billions of US-$, normalized to the year 2000 value (Munich Re) and (right) risk curves due to storms, floods, and earthquakes for the city of Cologne, Germany. The data considers losses at buildings and in the sectors private housing, commerce, and industry for the year 2000 (courtesy of G. Grünthal, 2004).
2 Research topics
Earthquake disasters are of course caused by the combination of strong ground shaking and buildings having low structural capacity, thus showing a poor performance during earthquake action and being unable to withstand the shaking without damages. Two main factors that therefore can turn an earthquake into a disaster are the vulnerability of (inadequately constructed) buildings, and unfavorable soil conditions beneath the building. The latter will amplify ground shaking effects and in some cases even contribute to liquefaction or sliding. The earthquakes in Mexico City 1985 and Armenia 1988 are prime examples of the importance of these two factors.
Many of the fast growing cities today are located on old lake beds and land fills with strong soil amplification potentials. The shaking contribution of these factors are important research targets in ICG Project 3, and Fig. 3 attempts to synthesize the overall earthquake vulnerability problem and some of the topics that are addressed in this project.
Figure 3. Sketch of some of the topics (inside the hatched circle) that are addressed in this project, while the neighboring research areas (outside the circle) are covered only more briefly.
2.1 Earthquake hazard and site amplification
NORSAR has been dealing with earthquake hazard-related problems for more than 30 years, both on the research side and as a partner for the construction industry. The efforts have partly been within earthquake hazard methodology, in particular probabilistic methods, but also to develop the underlying knowledge needed to perform sufficiently reliable analyses, concentrating on the following topics:
- earthquake hazard methodologies
- seismotectonics and earthquake source models
- ground-motion models (wave attenuation)
- site response (soil amplification)
Since the first two of these topics have been thoroughly covered in earlier works, the efforts within ICG Project 3 have been focused to some extent on the third topic but even more on the last one, as it will be shown in the subsequent sections. Fig. 4 illustrates some results from a study aimed at developing criteria for selecting and adjusting ground-motion models for specific target regions. It can be seen that the effect on the spectrum of the near-surface geology may be considerable, demonstrating the importance of detailed soil response and site characterization efforts. Below, a more detailed description of such investigations is presented.
Figure 4. VS30 (average shear-wave velocity of the soil in the uppermost 30 m) adjustment factors using `generic¿ rock models and VS30 and kappa (near-surface attenuation) adjustment factors. The coloured lines represent different ground motion (host) relations and the target kappa value is equal to 0.0125 s.
2.2 Seismic risk and loss analysis software (SELENA)
The single most comprehensive work towards earthquake risk calculation until today is condensed in HAZUS, a software system which was prepared by the Federal Emergency Management Agency (FEMA) for use in the United States. The basic approach behind this software is physical-analytical (and hybrid), and large resources had been used to define both capacity and fragility (vulnerability) curves for different building types and levels of design categories. From an engineering perspective this analytical approach is very attractive, however, it quickly becomes complex even for simple buildings, and a calibration to the damage records of historical earthquake events is necessary. Since HAZUS previously was an `inaccessible¿ software, NORSAR has developed a comparable stand-alone software that can be applied anywhere in the world, and which includes a logic tree-based weighting of input parameters that allows for the computation of confidence intervals. The open-source software package is called SELENA - Seismic Loss Estimation using a Logic Tree Approach.
Much of NORSARs recent work was concerned with the development of SELENA, which in turn establishes the basis for new research initiatives and applied projects. Since January 2007, SELENA is offered as open source software, available for download at http://www.norsar.no/seismology/selena.html, which we hope will attract the interest from new users. The SELENA software will now also be more widely used and distributed through research collaboration programs with India, Central America, and within different EU projects.
The SELENA/HAZUS approach is based on the 'capacity-spectrum method' since it combines the ground-motion input in terms of a response spectrum (spectral acceleration versus spectral displacement) with the building's specific capacity curve. The philosophy behind this is that any building is structurally damaged by its permanent horizontal displacement (and not by the acceleration itself). For each building and building type the inter-story drift (relative drift of the stories within a multi-story building) is a function of the applied lateral force that can be analytically determined and transformed into building capacity curves (capacity to withstand accelerations without permanent displacements). Building capacity curves naturally vary between different building types, and also between different regions, reflecting on building code regulations and local construction practice.
Figure 5. The principles of analytical risk assessment approaches (e.g. SELENA) are based on the crossover of physical capacity curves representing the nonlinear lateral displacement behaviour of the structure and the damped ground-motion displacement spectrum (top). Once the performance point is assigned damage probabilities for the five different damage states can be derived from the fragility curves (bottom).
Within the preparation of HAZUS, FEMA developed capacity curves for 36 building types (being representative for the United States) in four earthquake load regimes (reflecting the variation in building regulations as a function of time across the United States). These 144 capacity curves were analytically developed, but are adjusted so that empirical knowledge is incorporated in the curves whenever possible. The building capacity curve is described by three control points representing design, yield and ultimate capacity. Up to the yield point, the building¿s capacity curve is assumed to behave linearly (ideal elastic). From the yield point to the ultimate point, the capacity curve changes from an ideal elastic to a fully plastic state (curved form), and the curve is assumed to remain fully plastic past the ultimate point (linear form). A bilinear representation (two linear parts) is sometimes used to simplify the model. The vulnerability curves (also called fragility curves) are developed as lognormal probability distributions of damage from the capacity curves.
Using inventory data from Oslo (Norway) with 902 reinforced-concrete (RC) buildings from the district Grünerløkka, a simple classification of the buildings yields fragility curves as shown in Fig. 6, for Damage Limit States 1 to 3 (1: no damage, 2: slight damage, and 3: moderate damage).
Figure 6. Fragility curves for 902 reinforced-concrete buildings (1- to 3-story) in Oslo for Damage Limit States 1, 2, and 3 using a displacement-based approach.
2.4 Damage and loss estimation
Currently, SELENA computes seismic losses using the capacity spectrum method. The losses are computed as aggregates of the inflicted damages to each building type within a predefined area (a city quadrant, census tract or any user-specified area). The damage to the physical environment can then be converted to monetary damages as well as to estimates of casualties using empirical relations. The damages are presented in tabular form and any appropriate mapping or graphing software can be used to display the results in easy understandable figures.
2.5 Illustration of risk and loss results - RISe
One of the advantages of SELENA is its portability between platforms and its independence of graphic software. This independence can also be a disadvantage, however, if the potential user groups feel the lack of graphical interface as a deficiency.
In cooperation with INETER, one of NORSAR's long-term partners in Central America, a stand-alone software tool has been developed called RISe ¿ Risk Illustrator foer SELENA. RISe will allow a quick and easy illustration of the georeferenced input, inventory and output data of SELENA in GoogleEarth (GE) and will thus substitute the use of any commercial GIS program. Furthermore, the tool will help to demarcate the study areas in Google Earth and to generate the SELENA input files through an easy-to-understand and user-friendly program surface.
Different ways in order to illustrate the inventory data or risk results are implemented in RISe. Figures 7 and 8 shows two different ways of illustrating the damage probabilities of a particular model building type in the case study Oslo, Norway.
- The RISe software is today a stand-alone software solution which substitutes the use of a commercial GIS solution.
- It facilitates the preparation of SELENA input files.
- It allows the demarcation of study areas (census tracts) and illustration of inventory data and risk results in GoogleEarth.
Figure 7. Damage probability as illustrated by use of GoogleEarth. See legend for color code definitions.
Figure 8. Damage probability as illustrated by use of GoogleEarth. See legend for color code definitions.
2.6 Tsunami modelling
Up to now, initial seafloor displacements resulting from tsunami-triggering earthquakes are calculated using a simple, analytical method (e.g. Okada's method; Okada, 1992, BSSA) in most oceanographic models analysing tsunami inundation heights at neighbouring coastlines. Although analytical solutions are mathematically exact and inexpensive in terms of computation powers, they only allow for very simple model geometries and homogeneous material properties. Geist and Dmowska (1999, PAGEOPH) state that effects of heterogeneous slip patterns on the earthquake fault plane are important to include in the local tsunami wave field. The amount of slip for a given seismic moment depends not only on fault plane properties, e.g. the distribution of asperities, but also on the shear modulus of the surrounding material, which may vary with depth in subduction zones from 3 to 30 GPa (Bilek and Lay, 1999, Nature). Curvature of fault plane and seafloor topography may additionally influence the resulting displacement pattern. Such inhomogeneities can only be captured by a numerical model.
We use the FEM numerical method to deal with surface displacement computations resulting from slip on a fault. GeoFEST (Jet Propulsion Laboratory at California Institute of Technology; authors: G. Lyzenga and J. Parker) was chosen to be the most appropriate software solution for our task, since the code is capable to account for material inhomogeneities, a heterogeneous slip distribution on the earthquake fault plane and irregular fault geometry in 2-D or 3-D. Further, it is possible to include multiple faults, seafloor topography and friction on the fault.
Figure 9. Inhomogeneous FEM mesh used to calculate seafloor displacement resulting from slip on earthquake faults; mesh is subdivided to represent accretionary wedge, continental crust, oceanic plate and asthenosphere.
Figure 10. Influence of change of subduction angle from 0° to 90°; left: vertical displacement, right: horizontal displacement, light colour: steep subduction angle, dark colours: shallow subduction angle.
Figure 11. Influence of change fault plane bending; top: vertical displacement (red: bended fault plane, black: comparison to homogeneous model), bottom: sketch of bended fault plane.
2.7 Soil amplification studies
2.7.1 Numerical simulation of H/V-data
The work consists of using NGI¿s Green¿s functions software Laysac to simulate the steady-state horizontal and vertical motions on ground surface due to a distant source. The objective is then to use the simulated data to assess the performance/validity and the range of applicability of the H/V method for different soil profiles and especially for a range of stiffness contrast at the rock/soil interface. The results will also be compared with the associated transfer functions for the selected soil profiles.
2.7.2 Instrumental soil response studies
The amplification of seismic waves as they propagate through less consolidated sediments and soils is, as already mentioned, a major factor behind earthquake damages. The thick clay deposits in and around Oslo lends this region to such studies. Although Oslo is not a place with a significant seismic activity, a few earthquakes with noticeable intensity have occurred in the past, including a magnitude 5.4 earthquake in the outer Oslofjord in 1904. This earthquake caused masonry building walls to crack as well as chimneys and roof tiles to fall off the houses in the city of Oslo. The observed significant shaking (and the associated damages) was caused by wave amplification through the thick layers of sediments underlying Oslo.
To investigate this phenomenon in more detail we have conducted special studies both in 2004 and 2006, visiting 35 different sites within the city. The technique used insisted in recording ambient seismic noise and to infer the soil response on the basis of such data using the so-called Nakamura technique, based on ambient noise H/V (horizontal-to-vertical) ratios. The results (Fig. 12) are largely explaining the distribution of damages from the 1904 earthquake.
Figure 12. Spectral H/V ratio (median and the range of ± standard deviation) of site no. 28 (Oslo study). The peak at around 6 Hz indicates the first resonant frequency of the soil profile.
The simple Nakamura technique presented above is complemented by more elaborate methods which have also been tested and used under ICG Project 3, notably an array technique and a technique based on spectral analysis of surface waves (SASW). The former technique has been used in a conceptual study at Sogn Hagekoloni in Oslo, using two different array diameters. The data have been analyzed in terms of H/V ratios on individual stations, and a subsurface shear-wave velocity profile was established through joint analysis of the two array geometries (Fig. 13).
Figure 13. Ambient seismic noise recorded by two arrays with different aperture in order to conduct non-unique forward modelling of the velocity profile that matches the observed dispersion curves.
3 Related external projects
Several related projects are proceeding in 2008, including the EU projects LESSLOSS, NERIES, SAFER, and TRANSFER. In addition, MFA (UD) funded projects in Central America, Pakistan, and India. The coordination between ICG Project 3 and these projects will be very important in the coming years. Within NERIES, NORSAR is charged with coordinating the development of shake maps for Europe, and will also take part in the evaluation of risk estimation software and loss assessments.
Within SAFER, NORSAR is responsible for developing real time damage scenarios, where the city of Naples and possibly Bucharest will be used as targets. Another activity under the SAFER project is for NORSAR to develop rapid epicentre solutions based on array data processing. NORSAR also has a small component in TRANSFER with focus on tsunami-generating earthquake sources in the North Atlantic.
4 Plans for 2008
- Further development of the SELENA software with the main focus on distribution and application
- Expansion and adjustment of the RISe software
- Strengthening of structural modelling and nonlinear analysis capacities
- Further work on intra-plate ground-motion models (EU)
- Tsunami source modelling of fault dislocations (cooperation with ICG Project 10)
- Numerical simulation of H/V-data and further work on site amplification
- Cooperation and coordination with other projects (EU, MFA)
- Development of competence for multi-hazard and risk mitigation (earthquakes, tsunami, landslides)
- Visitors program (including students)
- Publications (papers and reports)
- 5 Publications 2007-2009
5.1 Papers with referee system
Bungum, H. (2007): Numerical modeling of fault activities. Computers & Geosciences 33: 808-820.
Molina S. and Lindholm, C. (2007): Estimation the confidence of earthquake damage scenarios: examples from a logic tree approach. Journal of Seismology 11(3): 399-310.
Lang, D.H., Molina, S., and Lindholm, C.D. (2008): Towards near-real-time damage estimation using a CSM-based tool for seismic risk assessment. Journal of Earthquake Engineering 12(S2), 199-210.
Prasad, J.S.R., Singh, Y., Kaynia, A.M., and Lindholm, C.D. (2008): Socio-Economic Clustering in Seismic Risk Assessment of Urban Housing Stock. Earthquake Spectra, accepted.
5.2 Other reviewed publications
Lang, D.H., R. Merlos, L. Holliday, and M. Lopez (2007): Vivienda de Bahareque in El Salvador. EERI World Housing Encyclopedia. Paper #141.
Lang, D.H., O. Flores, and L. Holliday (2007): Vivienda de adobe in Guatemala. EERI World Housing Encyclopedia. Paper #144.
Lang, D.H., A. Amador, L. Holliday, C. Romero, and A. Ugarte (2007): Vivienda de Minifalda in Nicaragua. EERI World Housing Encyclopedia. Paper #148.
Rautela, P., Joshi, G.Ch., Singh, Y. and Lang, D.H. (2008). Koti Banal Architecture of Uttarakhand, India. EERI World Housing Encyclopedia. Paper #150.
5.3 Papers in proceedings
Schwarz, J., Lang, D.H., Abrahamczyk, L., Bolleter, W., Savary, C., Bikce, M., Genes, M.C., and Kacin, S. (2007): Seismische Bauwerksinstrumentierung von mehrgeschossigen Stahlbeton-bauwerken - Ein Beitrag zum SERAMAR Projekt. D-A-CH Tagung Vienna/Austria, Sept. 2007.
Abrahamczyk, L., Schwarz, J., Lang, D.H., Leipold, M., Genes, M.C., Bikce, M. and Kacin, S. (2008). Building monitoring for seismic risk assessment (I): Instrumentation of RC frame structures as a part of the SERAMAR project, 14th World Conf. on Earthq. Eng. WCEE, Beijing/China, 2008.
Schwarz, J., Lang, D.H., Kaufmann, C., and Ende, C. (2007): Empirical ground-motion relations for Californian strong-motion data based on instrumental subsoil classification. Proceedings of the Ninth Canadian Conference on Earthquake Engineering, Ottawa, Ontario, Canada, 2007.
Schwarz, J., Langhammer, T., Leipold, M., Abrahamczyk, Kaufmann, Ch., Lang, D.H., and Riedel, S. (2008): Bewertung der Erdbebenverletzbarkeit eines Gebaeudebestandes in innerstaedtischen Grossraeumen - Phase 1 des SERAMAR Projektes. D-A-CH Mitteilungsblatt, Band 83, Sept. 2008, S2-S10.
Sigaran, C., Kaynia, A.M., and Hack, R. (2007). Soil stability under earthquakes: A sensitivity analysis. 4ICEGE. June 25-28. in review.
Haugen, E.D. and Kaynia, A.M. (2008). A comparative study of empirical models for landslide prediction using case histories. Proc. 14th World Conf. on Earthquake Engineering, Beijing, China, 12-17 Oct., Paper 04-02-0022.
5.4 Talks and poster presentations
Bungum, H., Faleide, J.I., Pettenati, F., Schweitzer, J. and Sirovich, L. (2008). The M 5.4 October 23, 1904, Oslofjord earthquake: Reanalysis based on instrumental and macroseismic data, 39th Nordic Seismology Seminar, Holmen Fjordhotell, Oslo, June 2008.
Lang, D.H. and Schwarz, J. (2007): The application of ambient seismic noise for engineering purposes. NATO Advanced Research Workshop 'Increasing Seismic Safety by Combining Engineering Technologies and seismological Data', Dubrovnik, Croatia, September 2007 (extended abstract).
Lang, D.H., Molina, S., and Lindholm, C.D. (2007): Towards near-real-time damage estimation using a CSM-based tool for seismic risk assessment. International Symposium on Earthquake Loss Estimation for Turkey (HAZTURK), September 2007, Istanbul, Turkey.
Lang, D.H., Molina Palacios, S., and Lindholm, C.D. (2007): The seismic risk and loss assessment tool SELENA and its applicability for (near-)real-time damage estimation. International workshop on seismicity and seismological observations of the Baltic Sea region and adjacent territories, September 10-12, 2007, Vilnius, Lithuania.
Lang, D.H., Molina, S., Lindholm, C.D. and Gutierrez, V. (2008). The seismic risk and loss assessment tool SELENA - Recent developments and applications, 39th Nordic Seismology Seminar, Holmen Fjordhotell, Oslo, June 2008.
Molina-Palacios, S., Galiana-Merino, J.J., Jiménez-Delgado, A., Zaragoza-Martínez, F., Jiménez, M.J., Gimeno-Nieves, E., Lang, D.H., and Lindholm, C.D. (2008). Seismic risk scenarios for urban areas of Alicante Province (southeast Spain), 14th World Conf. on Earthq. Eng. WCEE, Beijing/China, 2008.
Molina, S., Elias, C., Lang, D.H., Lindholm, C.D. (2008): Soil characteristics identification of urban areas in San Salvador (El Salvador) using H/V spectral ratio technique. 31st General Assembly of the European Seismological Commission ESC 2008, Crete, Greece, September 2008.
Roth, M. and Blikra, L.H. (2008). Seismic monitoring at the Åknes rock slope, Norway, European Geosciences Union General Assembly 2008, Vienna, Austria, April 2008.
5.5 Project reports
Harmandar, E. and H. Bungum (2007): Attenuation relationships for ShakeMap applications in Europe. Report for the NERIES (EU) project.
Lang, D.H., Lindholm, C.D., and Balan, S. (2007): Seismic risk and loss assessment for a selected study area in Bucharest, Romania. Technical report for the SAFER project, April 2007, 25 pp.
Molina, S., Lang, D.H., and Lindholm, C.D. (2007): SELENA v1.1 - User and Technical Manual v1.1, February 2007, 45 pp.
Molina, S., Lang, D.H., and Lindholm, C.D. (2007): SELENA v2.0 - User and Technical Manual v2.0, May 2007, 59 pp.
Molina, S., Lang, D.H., and Lindholm, C.D. (2007): SELENA v3.0 - User and Technical Manual v3.0, December 2007, 69 pp.
Molina, S., Lang, D.H., and Lindholm, C.D. (2008). SELENA v3.5.1 - User and Technical Manual v3.5.1, May 2008, 69 pp.
Molina, S., Lang, D.H., and Lindholm, C.D. (2008). SELENA v4.0 - User and Technical Manual v4.0, October 2008, 85 pp.
Lang, D.H., Gutierrez, V., and Lindholm, C.D. (2008). RISe v1.0 - User and Technical Manual v1.0, December 2008, 22 pp.
Amir M. Kaynia