United Nations Educational, Scientific and Cultural Organization and International Atomic Energy Agency

Transcription

1 Available at: IC/2009/075 United Nations Educational, Scientific and Cultural Organization and International Atomic Energy Agency THE ABDUS SALAM INTERNATIONAL CENTRE FOR THEORETICAL PHYSICS SEISMIC HAZARD EVALUATION FOR MAJOR CITIES IN MADAGASCAR Hoby N.T. Razafindrakoto, Gérard Rambolamanana Institute Observatory Geophysics of Antananarivo, University of Antananarivo, P.O. Box 3843, Antananarivo, Madagascar and The Abdus Salam International Centre for Theoretical Physics, Trieste, Italy and Giuliano F. Panza Dipartimento di Scienze della Terra, Università di Trieste, Trieste, Italy and The Abdus Salam International Centre for Theoretical Physics, Trieste, Italy. Abstract The seismic hazard in some areas in Madagascar has been assessed at regional scale in terms of peak ground motion values (displacement, velocity, acceleration) and their periods, following the Neodeterministic approach, based on the computation of realistic synthetic seismograms. The main data input integrates all available tectonic, seismicity and structural model information. The largest peak values are 1.6cm/s for the velocity, 0.03g for the acceleration and more than 0.5cm for the displacement. These values are consistent within a range of macroseismic intensity from VI to VII MCS, and indicate that relatively simple prevention measures and retrofitting actions may guarantee a high safety level and a well sustainable development. MIRAMARE - TRIESTE September 2009

2 INTRODUCTION Although Madagascar belongs to an intraplate context, it has been affected by some destructive earthquakes such as the events of November 3, 1897; March 29, 1943 and April 21, 1991 (Poisson, 1924; 1930; Bertil et al., 1998). For the 1991 event, good observations are available as well as official estimates of damages. Since 1991, considerable effort has been focussed on the determination of the different parameters needed to achieve the seismic hazard assessment of Madagascar: seismic source study (Andriamarondranto, 1995; Barimalala Rondrotiana, 2006) and velocity structure determination (Rambolamanana, 1999; Rambolamanana et al., 1997). This assessment is necessary to protect the population and man-made structures. A reliable and practically useful assessment requires, not only the identification of the safest and most suitable areas for urban development, but also the definition of the seismic input that is going to affect a given building, for the optimization of its design. Seismic hazard assessment can be performed in various ways, following a probabilistic (PSHA) or a deterministic (DSHA) approach (Reiter, 1990). Both depend on the definition of the seismicity of the area, the geological and geotechnical conditions and finally, the estimation of the largest magnitude of future earthquakes. But they have differences, advantages and disadvantages that often make the use of one method preferable to the other. McGuire (2001) discussed the factors that influence the choice of the seismic hazard assessment procedure including the decision to be made, the seismic environment, the available input data and the scope of the assessment. The probabilistic approach is based on rough assumptions and models (e.g. recurrence and attenuation laws), and cannot take into account some of the most important ingredients of hazard- like rupture process, directivity and site effects. This is evidenced by the comparison (table 1) of recent recordings with the values predicted by the probabilistic methods (e.g. Kobe, ; Bhuj, ; Boumerdes, ; Bam, ; Sichuan, ). Moreover, the mathematical model of PSHA is inaccurate and introduces systematic errors in the calculation process that contribute to obtain unrealistic results (Klügel, 2007). An accurate and physically sound hazard assessment has been made in this study following a neo-deterministic (NDSHA) approach, which allows to give a realistic description of the seismic ground motion due to an earthquake of given distance and magnitude (e.g. Panza et al., 2001). METHOD NDSHA (Zuccolo et al., 2008; Panza et al., 1996; Panza et al., 2001) can be used as a starting point for the development of an integrated approach that combines the advantages of the probabilistic and deterministic methods and permits to assess the seismic hazard at various scales: here the regional scale is considered. The approach is based on modelling techniques that have been developed from the knowledge of the seismic source and propagation processes. The scheme of the procedure is shown in figure 1. When the seismicity, the source mechanisms, the structural models and the observation points are defined, a set of synthetic seismograms is generated by means of the modal summation technique (Panza, 1985; Panza and Suhadolc, 1987; Florsch et al., 1991). The horizontal P-SV and SH components of motion are computed 2

3 and rotated to the N-S and E-W directions after which their vector sum is calculated. The observation points are considered to be at the nodes of a grid with step 0.2 used to discertize the investigated area. All of the generated seismograms are then examined to extract the peak ground displacement (PGD), velocity (PGV), acceleration (PGA) and the corresponding periods, which are the simplest parameters relevant for seismic engineering. GMT software (Wessel and Smith, 1991) were used to prepare some of the figures. This procedure, very fruitful since it allows to obtain a realistic estimate of hazard without having to wait for a strong event to occur, has already been applied successfully to many areas at different scales worldwide such as Italy (Panza et al., 2001), India (Parvez et al., 2003) and Bucharest (Cioflan et al., 2004); for a review see Panza et al. (2002). INPUT DATA To assess the seismic hazard of Madagascar, we have used the database of the earthquakes with magnitude greater than 3.5 detected by the local seismic network in the time interval from 1975 to 2009 (figure 2). Since such catalogue is both incomplete and affected by errors, a smoothing procedure is applied to generalize the distribution of earthquake sources (Panza et al., 1996). Through the NDSHA approach, only the events within the seismogenic zones are considered. Rakotomalala has recently defined 29 zones for Madagascar (Rakotomalala, 2007; Rakotomalala et al., 2009). But in the present paper, only 8 seismogenic zones have been used by combining some of the previous zones and neglecting the others, where the focal mechanism is unknown. This decision limits the validity of the results to just a part of the country, but we believe that this decision causes a minor underestimation of hazard since the lack of information about focal mechanism can be a natural indication of the relatively small size of the events involved. Nevertheless, if for some of the neglected areas there is some special and urgent need to assess the seismic hazard, this can be easily done by NDSHA, by means of intensive parametric studies involving the parameters describing the source mechanism. The NDSHA approach can treat the seismic source as a point as well as a finite dimensions object. But in this study, we limit to a point source since the available information does not allow to estimate the size of the fault and related details. Such point source is scaled for their dimensions using the relatively simple spectral scaling laws by Gusev (1983). The focal mechanisms defined for each seismogenic zones (figure 3) are taken from CMT Harvard. The structural models of the lithosphere beneath Madagascar are given in table 2.The thickness, the density and the P and S waves velocity of each layer are taken from the average model proposed by Rambolamanana (1997). RESULTS Figure 4, where the distribution, within the seismogenic zones, of the smoothed magnitudes is shown, represents the distribution of the point source events used as scenario earthquakes to compute the synthetic 3

4 seismograms and therefore the ground motion field. From these computations, using NDSHA procedure (Panza et al., 2001), seismic hazard maps of selected areas in Madagascar have been obtained in terms of peak values and the associated periods for PGA (figure 5), PGV (figure 6), PGD (figure 7). The lack of information in some regions of Madagascar, such as in the Northern and Southern parts, is due to the lack of definition of seismogenic zones. A possibility to overcome this problem is offered by the application of the morphostructural zonation and the pattern recognition algorithm to the identification of earthquake prone (Gorshkov et al., 1991b; 2000; 2004). The maximum PGA is about 0.03g and it matches with the result obtained by Giardini et al. (1999) for Madagascar, in which the PSHA estimate of PGA for a return period of 475 years varies in the range of 0.02g to 0.04g. This value is rather low when compared with values obtained in other intraplate seismicity areas, like UK, where the PSHA estimate of PGA varies in the range of 0.05g to 0.25g (Musson, 1997). For the case of Madagascar, the maximum values (PGA between 0.02g and 0.03g) occur in the region of Famoizankova (18 S, 46 E) and Ifanadiana (20.5 S, 47.5 E). The PGA value in the region of Famoizankova is controlled by the April 21, 1991 event, for which Rambolamanana and Ratsimbazafy (1991) have reported the damaging effects. In fact, earthquake damage is quite different from one fokontany (locality with less than 30 houses) to another: cracks appeared in about 90% of people's houses in Ambohimiarina, about 84% in Kiranomena, 33% in Beanana and 30% in Fenoarivobe. These data represent the only empirical evidence that can be used to validate the results of the theoretical modelling. From figure 8, which is a simple blow-up of a part of figure 6, we can conclude that the available observed damage distribution matches with the computed peak values. The largest PGA is computed in the region of Kiranomena and Ambohimiarina (about 90% of the houses were cracked), and is in the range from 0.02 to 0.03g. These two areas correspond to the zone where the highest intensity has been experienced during the April 21, 1991 earthquake. Accordingly, with the conversion (table 3, table 4) given by Panza et al. (1997) the MCS macroseismic intensity can be estimated as large as VII. In the other two fokontany, Beanana and Fenoarivobe (about 30% of the houses were cracked), the intensity is evidently less than in the previous ones and since the PGA is between 0.01 and 0.02g, can be estimated as large as VI (MCS). The results of our study, shown in figure 6, permit to infer the PGA in the main industrial and large cities in Madagascar where most of the sensitive constructions and structures are located such as Antananarivo, Antsirabe, Fianarantsoa and Tamatave. In Antananarivo, for example, the computed ground motion (PGD, PGV, PGA) are about 0.1cm, 0.5cm/s and 0.005g, respectively, and therefore the seismic hazard in this area can be considered practically negligible. On the other side, the computation in the area of Antsirabe leads to PGD, PGV and PGA values in the range from 0.1 to 0.5cm, from 1 to 1.6cm/s and from 0.01 to 0.02g respectively, that correspond to a macroseismic intensity as large as VI (MCS) (see table 3 and table 4). The period at which the peak values occur is also an important parameter for engineers in order to check the 4

5 response of the buildings. Actually, one structure is more hazardous than others if the computed period coincides with or is close to the natural frequency of the building. For example, figure 7 shows that in the region of Ifanadiana, the periods in the range from 1 to 2 s are predominant, which correspond to the natural frequency of flexible structures such as 10-story buildings. These results are very useful especially in terms of pre-disaster management. Since it is not feasible to control nature and to stop the development of natural phenomena like earthquakes, seismic hazard assessment is fundamentally a predictive and preventive effort that has been made to reduce the threat of earthquakes. The existing law in Madagascar, in terms of civil construction does not involve seismic hazard. Otherwise, to effectively mitigate the impact of future earthquakes, engineers should refer their building conception with respect to the PGA and the corresponding periods for a given area. CONCLUSION The estimation of the seismic hazard of selected regions in Madagascar has been obtained using the NDSHA based on the computation of realistic seismograms. This is very useful for earthquake preparedness especially in the most populated areas. The seismic hazard for Madagascar is, in general, below the damage level with the exception of some areas such as Famoizankova, Antsirabe, Ifanadiana where the intensity can be as high as VII (MCS) and cause severe cracks to standard houses. The largest peak values corresponding to those areas are 1.6 cm/s for the velocity, 0.03g for the acceleration and more than 0.5 cm for the displacement. The obtained results of the present study should be used as a starting point for risk study in Madagascar, and it can be extended to the whole island, when the relevant data will become available. ACKNOWLEDGMENTS This work was carried out at the Institute and Observatory of Geophysics in Antananarivo (I.O.G.A.), in collaboration with ESP-SAND group of ICTP, Trieste, Italy. We would like to express our gratitude to the personnel at IOGA and the Department of Earth Sciences of the University of Trieste. Our special thanks go to Fabio Romanelli and Franco Vaccari for their assistance. We acknowledge also the Département d'analyse et Surveillance de l'environnement (DASE/LDG), France for their collaboration. Funding from ICTP-STEP Programme is also gratefully acknowledged. 5

6 REFERENCES Andriamarondranto, R.J.V., (1995). Determination des paramètres aux sources dans les zones sismiques de Madagascar, These, Universite d Antananarivo Barimalala, R., (2006). Solutions de certains mécanismes au foyer régional de Madagascar, DEA, Université d Antananarivo Bertil, D., Regnoult, J.M., (1998). Seismotectonics of Madagascar, Tectonics 294, Cioflan, C.O., Apostol, B.F., Moldoveanu, C.L., panza, G.F., Marmureanu, Gh., (2004). Deterministic approach for the seismic microzonation of Bucharest, Pure and Appl. Geoph.161, Florsch, N., Fäh, D., Suhadolc, P. and Panza, G.F., (1991). Complete synthetic seismograms for highfrequency multimode SH-waves, Pure and Appl. Geoph. 136, Giardini, D., Grünthal, G., Shedlock, K., Zhang, P., (1999), The GSHAP Global Seismic Hazard Map Gorshkov A., Filimonov M., Tumarkin A., Rantsman E., Zhidkov,M., (1991b). How morphostructural zoning may contribute to the seismic one: the Lesser Caucasus. Proceedings of the fourth international conference on seismic zoning. August, 25 th 29 th. Stanford University, USA. Vol.2, Gorshkov A., Kuznetsov I., Panza G., Soloviev A. (2000). Identification of future earthquake sources in the Carpatho-Balkan orogenic belt using morphostructural criteria. Pure and Appl. Geophys., 157, Gorshkov A., Panza G., Soloviev A., Abdelkrim A. (2004). Identification of seismogenic nodes in the Alps and Dinarides., Boll.Soc.Geol.It., 3-18, 4 ff., 9 tabb Gusev, A.A., (1983). Descriptive statistical model of earthquake source radiation and its application to an estimation of short period strong motion. Geophysical Journal of the Royal Astronomical Society 74, Klügel, J.-U., (2007). Error inflation in probabilistic seismic hazard analysis. Engineering Geology 90, McGuire, R.K., (2001). Deterministic vs. probabilistic earthquake hazard and risks. Soil Dynamics and Earthquake Engineering 21 (2001), pp Musson, R.M.W., (1997). Seismic Hazard studies in the U.K.:Source specification Problems of Intraplate Seismicity, Natural Hazard 15, Panza, G.F., (1985). Synthetic seismograms: the Rayleigh waves modal summation. J. Geophys., 58, pp Panza, G.F., Cazzaro, R. and Vaccari, F., (1997). Correlation between macroseismic intensities and seismic ground motion parameters. Annali di Geofisica, Vol. XL, N. 5, Panza, G.F., Alvarez, L., Aoudia, A., Ayadi, A., Benhallou, H., Benouar, D., Bus, Z., Chen, Y., Cioflan, C., Ding, Z., El-Sayed, A., Garcia, J., Garofalo, B., Gorshkov, A., Gribovszki, K., Harbi, A., Hatzidimitriou, P., Herak, M., Kouteva, M., Kuznetzov, I., Lokmer, I., Maouche, S., Marmureanu, G., Matova, M., Natale, M., Nunziata, C., Parvez, I.A., Paskaleva, I., Pico, R., Radulian, M., Romanelli, F., Soloviev, A., Suhadolc, P., Szeidovitz, G., Triantafyllidis, P. and Vaccari, F., (2002). Realistic modeling of seismic input for megacities and large urban areas (the UNESCO/IUGS/IGCP project 414). Episodes, 25,

7 Panza, G.F., Cazzaro, R. and Vaccari, F., (1997). Correlation between macroseismic intensities and seismic ground motion parameters. Annali di Geofisica, 40, Panza, G.F., Romanelli, F., Vaccari, F., (2001). Seismic wave propagation in laterally heterogeneous anelastic media: theory and applications to the seismic zonation, Advances in Geophysics, Academic press, 43, 1-95 Panza, G.F., Suhadolc, P. (1987). Complete strong motion synthetics, In: Seismic Strong Motion Synthetics, Computational Techniques 4, B.A. Bolt (Editor), Academic Press, Orlando, Panza, G.F., Vaccari, F., Costa, G., Suhadolc, P., Fäh, D. (1996). Seismic input modelling for zoning and microzoning, Earthquake Spectra, 12, Parvez, I. A., Vaccari, F., Panza, G. F., (2003). A deterministic seismic hazard map of India and adjacent areas. Geophysical journal international (2003), 155, Poisson, R.P., (1924). Tremblements de Terre de Tananarive: Bulletin Economique de Madagascar, Tananarive Poisson, R.P., (1930). Tremblements de Terre Malgaches de 1897 à Bulletin Economique de Madagascar, Tananarive Rakotomalala, M.S., (2007). Etude de la sismicité de Madagascar en relation avec la tectonique, DEA, Université d Antananarivo Rakotomalala, M., Razafindrakoto, H., Rambolamanana, G., (2009). seismicity and tectonics of Madagascar, IASPEI General Assembly 2009 Rakotondrainibe, (1977). Contribution à l'étude de la sismicité de Madagascar. Thèse, Univ. Antananarivo Rambolamanana, G., Ratsimbazafy, J.B., (1991). Le séisme du 21 Avril 1991 localisé dans la région de Famoizankova, Rapport à l'académie Malgache (1991) Rambolamanana, G., Suhadolc, P., Panza, G.F., (1997). Simultaneous Inversion of Hypocentral Parameters and Structure Velocity of the Central Region of Madagascar as a Premise for the Mitigation of Seismic Hazard in Antananarivo, Pure appl. Geophys. 149, Rambolamanana, G., (1999). Modelisation de la partie Centrale de Madagascar par la Sismologie, These, Universite d Antananarivo Reiter, L., (1990). Earthquake Hazard Analysis, Colombia University Press Wessel, P., and Smith, W.H.F. (1991). Free software helps map and display data, EOS Trans. AGU, 72, 441 Zuccolo, E., et al. (2008). Neo-deterministic definition of seismic input for residential seismically isolated buildings, Engineering Geology (2008), doi: /j.enggeo

8 Table 1. Comparison between expected and observed PGA values for a few recent strong events. Expected PGA values refer to a probability of excedence of 10% in 50 years (return period of 475 years). When available, DGA values predicted by NDSHA are given. Earthquake PGA expected (g) PGA observed (g) DGA predicted (g) Kobe (Japan, ) Gujarat (India, ) Boumerdes(Algeria, ) Bam (Iran, ) W-Sichuan (China, ) >0.8 (Shakemap) Table 2. Structural model. Thk (km) rho Vp (km/s) Vs (km/s) Qp Qs 10 2,81 5,98 3, ,87 6,38 3, ,90 6,63 3, ,30 8,00 4, ,30 8,10 4, Table 3. ING NT4.1.1 (MCS) (horizontal components). Intensity PGD(cm) PGV(cm/s) DGA (g) V VI VII VIII IX X XI Table 4. ISG NT4.1.1 (MCS) (horizontal components). 8

9 Intensity PGD(cm) PGV(cm/s) DGA (g) VI VII VIII IX

10 Figure 1: Scheme of the neo-deterministic procedure for seismic hazard assessment at regional scale (Panza et al., 2001) 10

11 Figure 2: Distribution of earthquakes with a magnitude greater than 3.5 that occured in Madagascar between 1975 and

12 Figure 3: Focal mechanism associated with each seismogenic zone 12

13 Figure 4: Smoothed magnitude distribution for the cells belonging to the seismogenic zones 13

14 Figure 5: Horizontal Peak Ground Displacement distribution and Period in seconds of its maximum 14

15 Figure 6: Horizontal Peak Ground Velocity distribution and Period in seconds of its maximum 15

16 Figure 7: Horizontal Peak Ground Acceleration distribution and Period in seconds of its maximum 16

17 Figure 8: Horizontal Peak Ground Acceleration distribution of some areas in Madagascar 17