6th Framework Programme Priority 1.1.6.3 – Global Change and Ecosystems

1. 6th Framework Programme Priority 1.1.6.3 – Global Change and Ecosystems The LESSLOSS Project Risk Mitigation for Earthquakes and Landslide Simulating Earthquake Scenarios in the European Project LESSLOSS: the case of the Metropolitan Area of Lisbon (MAL) INGV-LNEC Team ZonnoG.1, Carvalho A.2,Franceschina G.1, Campos Costa A.2, Coelho E.2, Akinci A.1, Cultrera G.1, Pacor F.1, Pessina V.1 and Cocco M.1 1)INGV - Istituto Nazionale di Geofisica e Vulcanologia, Italy 2)LNEC - Laboratório Nacional de Engenharia Civil, Portugal

2. SIMULATING EARTHQUAKE SCENARIOS IN THE EUROPEAN PROJECT LESSOSS Within the framework of the LESSLOSS – Risk Mitigation for Earthquakes and Landslides ground motion scenarios are computed on the basis of the most probable 50 and 500 years Return Period event defined by specific location and magnitude for three urban areas: Lisbon (Portugal), Thessaloniki (Greece) and Istanbul (Turkey) Seismic Hazard Map of the European-Mediterranean region, in terms of peak ground acceleration at a 10% probability of exceedance in 50 years [from Jiménez, Giardini and Gruenthal, 2003]. Istanbul Thessaloniki Lisbon

3. The SP 10 “Disaster scenarios predictions and loss modeling for urban areas” This contribution “Simulating Earthquake Scenarios in the European Project LESSLOSS: the case of the Metropolitan Area of Lisbon (MAL)”is a part of Sub-Project 10 of the LESSLOSS Project, “Disaster scenarios predictions and loss modeling for urban areas”; The overall aim of SP 10 is: “to create a methodology, based on state-of-the-art loss modeling software, to provide strong, quantified statements about the benefits and costs of a range of possible mitigation actions, to support decision-making by city and regional authorities for seismic risk mitigation strategies”

4. Estimation of ground shaking using Deterministic Seismic Hazard Analysis (DSHA) Earthquake scenario = for a given site, the ground shaking level (in terms of PGA, PGV, SA and time series) is computed with a specific finite-fault method and with given hypothesis regarding the reference earthquake. • Finite-fault effects and directivity are important because of the relative nearness of seismic source; • The urban level of losses scenario ask for high resolution of ground motion input to match with the complexity of geotechnical characterization, vulnerability data and exposure factors; • When applied to critical structures DSHA provides a straightforward framework for evaluating the worst-case ground motions.

5. A general framework for the HAZARD STUDY procedurein the case of Lisbon OUTPUT TOOLS INPUT Probabilistic maps PSHA Historical and seismotectonical considerations Deaggregation analysis Choice of seismic source DSMRSSIM FINSIM Properties of the source and the crustal medium Ground motion at bedrock Geotechnical data LNECloss System Surface shaking map

6. Analysis of the active faults for the selection ofthe reference earthquakes O F F SH O R E S O U R C E S Active faults in SW Iberia (modified from Zitellini et al., 2005), considerable as probable sources for the 1755 earthquake. MPF – Marquês de Pombal Fault, PSF – Pereira de Sousa Fault, HSF – Horseshoe Fault, GBF – Guadalquivir Bank Faults. Red star represents location of 1969 earthquake. INLAND SOURCES Neotectonic of the Tagus Valley Region (modified from Vilanova and Fonseca, 2004). VF- Vila Franca Fault, ArR- Arrábida range, AF-Alcochete fault. Double dot represents location of 1909, no bold represents location of the 1531, Benavente earthquakes.

7. DE-AGGREGATION of PSHA for the Metropolian Area of Lisbon The seismic action scenarios were defined on the basis of the obtained modal values derived from the last recent study (Sousa, 2006) on 3D de-aggregation analyses in M and (X, Y) (magnitude and coordinates of bin source)

8. Reference earthquakes for given Return Period using the de-aggregation of the PSHA The reference earthquakes with Return Period of 50 and 500 years have been used to evaluate scenarios as input for loss modeling in the Metropolitan Area of Lisbon The reference earthquakes with Return Period of 200 years have been used to evaluate and compare scenarioswith the different methods and to do the treatment of the uncertainty

9. Test points LTVF M 5.7 LTVF M 4.4 MPTF M 7.9 MPTF M 7.6 Parishes of MAL Metropolitan Area of Lisbon and the selected Reference Earthquakes (for different RP) Metropolitan Area of Lisbon

10. Simulating Earthquake Scenarios using Finite-Fault Model A General Scheme To Evaluate Ground Motion At The Site Using a Finite Fault

11. Simulating Earthquake Scenarios using Finite-Fault Model The Finite-Fault Model Parameters Finite-fault simulations require informationonthe fault-plane geometry (length, width, strike and dip), the source parameters (seismic moment, slip distribution, stress drop, nucleation point, rupture velocity, etc.), the crustal properties of the region (geometrical spreading coefficient, quality factor, etc.) and the site-specific soil response. A dataset of digital acceleration records obtained from the Portuguese accelerometer network at hard-rock sites was employed to calibrate specific simulation parameters

12. Portugese digital network to calibrate the crustal properties of the region Portugese digital network (above) and epicenter locations of some recorded earthquakes (right)

13. THE NUMERICAL APPROACHES USED IN THE EUROPEAN PROJECT LESSLOSS • In the European Project LESSLOSS SP 10 we evaluate shaking scenarios using the following Finite-fault approaches: • Deterministic-Stochastic simulation Method, DSM (Pacor et al., 2005) (used for three cities); • Non-stationary stochastic Finite fault simulation Method, RSSIM (Carvalho et al., 2004) (used only for MAL). FURTHERMORE to compare and to calibrate the methods we use also: FINSIM (Beresnev and Atkinson, 1998), a stochastic finite-fault method andEXSIM (Motazedian and Atkinson,2005): a stochastic finite-fault method with “the dynamic corner frequency” option.

14. N DSM : Deterministic Stochastic Method (Pacor et al., 2005) P1 : directive site P2: anti-directive site The methods DSM, FINSIM and RSSIM are all Finite-Fault approaches that are based on a modification of the Point-Source-Stochastic-Methodof Boore [2003] P2 The DSM computes two horizontal components of motion and reproduces very well the directive effects P1 km

15. RSSIM : Non-stationary stochastic finite fault simulation method (Carvalho et al., 2004) The RSSIM method is customized for engineering applications it works inside the LNECLoss system. It is a spectral approach. The earthquake scenarios for each parish are defined by the computation of the Power Spectra Density Function (PSDF) of surface acceleration with due consideration of local site effects. A new stochastic procedure, after Serra Bilé and Caldeira, 1998

16. TREATMENT OF UNCERTAINTY In order to evaluate the variability depending on the uncertainties of input parameters a sensitivity analysis is done varying the following input parameters: • Different velocities of rupture (Vr) • Different nucleation points: bilateral propagation (NP = 2) and two others as unilateral direction with the nucleation point at the bottom corners (NP =1 and NP = 3) • homogeneous slip distribution model for the Lower Tagus Valley Fault (LTVF)M 5.7 with Return Period (RP) 200 years • Different different slip distribution model for the Marques Pombal Thrust Fault (MPTF)M 7.6 with Return Period (RP) 200 years LTVFM 5.7

17. TREATMENT OF UNCERTAINTY. DSM method: changing rupture nucleation points Lower Tagus Valley Fault (LTVF) M 5.7 (RP 200 yr) at the BEDROCK

18. TREATMENT OF UNCERTAINTY. DSM method: changing rupture velocity and nucleation points Lower Tagus Valley Fault (LTVF) M 5.7 (RP 200 yr) S021 S174 S137 S250 S268 Vr = 2.7 km/sec; NP = 2 (bilateral case) at the BEDROCK

19. TREATMENT OF UNCERTAINTY. DSM method: changing rupture velocity and nucleation points Lower Tagus Valley Fault (LTVF) M 5.7 (RP 200 yr) S021 S174 S137 S250 S268 Vr = 2.7 km/sec; NP = 1 (directive case) at the BEDROCK

20. Vr = 2.7 km/sec; NP = 3 (directive case) TREATMENT OF UNCERTAINTY. DSM method: changing rupture velocity and nucleation points Lower Tagus Valley Fault (LTVF) M 5.7(RP 200 yr) S021 S174 S174 S137 S137 S250 S250 S268 at the BEDROCK

21. 16 DIFFERENT MODELS NUCLEATION POINTS SLIP 1 SLIP 2 SLIP 3 SLIP RANDOM 3 2 1 R R 1 2 3 1 3 3 R 1 3 2 R TREATMENT OF UNCERTAINTY: changing slip model and nucleation points SLIP DISTRIBUTIONS and NUCLEATION POINTS(Gaussian distribution centered on nucleation points) 1 2 3 1

22. Shaking at the BEDROCK 16 DIFFERENT MODELS S174 NUCLEATION POINTS SLIP 1 SLIP 2 SLIP 3 SLIP RANDOM 3 2 1 R R 1 2 3 1 3 3 R 1 3 2 R TREATMENT OF UNCERTAINTY. FINSIM method: changing slip model and nucleation points MPTF M 7.6 RP 200 yr

23. TREATMENT OF UNCERTAINTY. FINSIM method: changing slip model and nucleation points Inferred variability could be treated in a statistical way but for the loss modeling in MAL “the worst-case” was adopted with purpose of generating precautionary scenarios only The yellow bars are the total frequency histogram of 480 times series (16 models x 30 trials) at the site S174 in case of the MPTF M 7.6

24. The shaking at the surface was evaluated using the LNECloss system The site effects are evaluated by means of an equivalent stochastic non-linear one-dimensional ground response analysis of stratified soil profile units properly designed for the Metropolitan Area of Lisbon

25. Lower Tagus Valley Fault (LTVF)Mw 4.4 PGA scenario (Return Period 50 years) used to evaluate loss modeling in MAL (Boomer attenuation law) at the BEDROCK at the SURFACE

26. Marques Pombal Thrust Fault (MPTF) Mw 7.9 PGA scenario (Return Period 500 years) used to evaluate loss modeling in MAL (attenuation law using RSSIM method) at the BEDROCK at the SURFACE

27. RESULTS and COMMENTS • In this study we simulate the ground motion shaking (at the bedrock and surface levels) for the Metropolitan Area of Lisbon (MAL) with two proposed scenarios: Lower Tagus Valley Fault (LTVF) Mw 4.4 (RP 50 years) and Marques Pombal Thrust Fault (MPTF) Mw 7.9 (RP 500 years); • The ground motion shaking have been computed first at the bedrock sites. To include the local site effects, available microzonation studies have been used to characterize the site amplification. In the case of Lisbon, the characterization of local soil effects is taken into account, using the LNECloss system, computing the Power Spectra Density Function (PSDF) at the surface level; • A treatment uncertainty was carried out for different methodologies (DSM, RSSIM and FINSIM) and the variability of ground motion was studied. The variability of ground motion was inferred from a set of fault ruptures scenarios incorporating different nucleation points, different ruptures velocities and slip distribution.

28. RESULTS and COMMENTS • Following the purposes of the SP10 we have worked with existing methods and the GIS-based software as the LNECLoss system (Campos Costa et al., 2002) that comprises several modules to perform seismic risk analyses. The only two used modules concerning this study are the Bedrock Seismic Input and Local Soil Effects evaluation • An improvement of the LNECloss system could be possible incorporating the new approaches and all knowledge on sensitivity studies gained during the European Project LESSLOSS.