PROBABILISTIC SEISMIC HAZARD ASSESSMENT IN ROMANIA: APPLICATION FOR CRUSTAL SEISMIC ACTIVE ZONES

PROBABILISTIC SEISMIC HAZARD ASSESSMENT IN ROMANIA: APPLICATION FOR CRUSTAL SEISMIC ACTIVE ZONES I. A. MOLDOVAN, E. POPESCU, A. CONSTANTIN National Institute for Earth Physics, C.P. MG-2, Calugareni 12, 077125, Magurele, Ilfov, Romania E-mail: iren@infp.ro Received August 8, 2007 The paper is adverted to a complex activity of research concerning the settlement of the crustal seismic hazard in Romania for risk studies. The seismic risk assessments are very important to specialized organizations, as these studies reveal (physically meaning) the neuralgic points of certain sites that contain constructions, representing real centers of potential disasters when stroked by natural catastrophes and having large socio-economical impact. To forestall such catastrophes, two categories of measures are necessary: (i) achievement of studies regarding the seismicity and evolution of the seismicity within the area, in order to detect abnormalities related to temporal evolution of the systems, abnormalities that can be considered as forerunners of major earthquakes; (ii) hazard and seismic risk assessment for strategic sites and the dissemination of the information in the decision media. The analysis that we propose implies: (1) geometrical definition of all seismic sources affecting Romania, (2) estimation of the maximum possible magnitude, (3) estimation of the frequency magnitude relationship, (4) estimation of the attenuation law and, finally, (5) computing PSH. Key words: crustal seismicity, seismic hazard, probabilistic approach. 1. INTRODUCTION The seismic hazard assessment in dense-populated geographical regions and subsequently the design of the strategic objectives (dams, nuclear power plants, etc.) are based on the knowledge of the seismicity parameters of the seismogenic sources which can generate ground motion amplitudes above the minimum level considered risky at the specific site and the way the seismic waves propagate between the focus and the site. Extremely vulnerable objectives, like large cities, hidroenergetic dams or nuclear power plants, are present all arround Romania, and not only in the Vrancea intermediate earthquakes action zone. The best example is the western part of Romania, that is not affected by Vrancea intemediate earthquakes and where the crustal seismicity is high. In this part of the country are cities like Timisoara, Arad and Oradea and the Portile de Fier I and II hidroenergetic Rom. Journ. Phys., Vol. 53, Nos. 3 4, P. 575 591, Bucharest, 2008

576 I. A. Moldovan, E. Popescu, A. Constantin 2 dams. Therefore, this region was primarily considered in all the studies of seismic hazard in Balkan and Circum-Pannonian regions (see for example the topical volumes Vrancea Earthquakes: Tectonics, Hazard and Risk Mitigation in Kluwer Academic Press, 1999 and Seismic Hazard of the Circum-Pannonian Region in Pure and Applied Geophysics, vol. 157, 2000). The purpose of this paper is to provide a complete set of information required for a probabilistic assessment of the seismic hazard in Romania relative to the crustal sources. The analysis that we propose implies: (1) geometrical definition of all seismic sources affecting Romania, (2) estimation of the maximum possible magnitude, (3) estimation of the frequency magnitude relationship, (4) estimation of the attenuation law and, finally, (5) computing PSH with the algorithm of [1]. 2. SEISMIC SOURCES CHARACTERISTICS The first step in the determination of probabilistic crustal hazard consists in defining the seisogenic sources. It is necessary to point out and to delimit the seismic areas from the Romanian territory. The seismogenic crustal sources (Fig. 1) that affect the Romanian territory are: Vrancea crustal source (VRN), Barlad Depression zone (BD), Predobrogean Depression source (PD), Intramoesian fault-shabla-dulovo source (IMF-DUL-SH), Fagaras-Campulung- Sinaia crustal sources (CMP = FG + CP + SI), Transilvanian Depression (TD), Banat crustal source (BAN), Danubian crustal earthquakes (DAN) and IBAR zone, Crisana-Maramures sources (CM = CMS1 + CMS2). 2.1. VRANCEA CRUSTAL SOURCE The seismic activity in Vrancea in the crustal domain (VN) is located in front of the Southeastern Carpathians arc, spread over a stripe area delimited to the north by the Peceneaga-Camena fault and to the south by the Intramoesian fault (Fig. 1). The seismicity is more diffuse than for the subcrustal source and consists only in moderate-magnitude earthquakes (M w 5.5) generated in clusters localized in the eastern part (seismic sequences of Râmnicu Sarat area) and northern part (swarms in Vrâncioaia area). The catalog contains a single earthquake of M w = 5.9 occurred on March 1, 1894, with magnitude estimated from historical information (possibly overestimated). The rate of the seismic moment release, Mo = 5.3 1015 Nm/year, is four order of magnitude less than the moment rate characteristic for the Vrancea intermediate-depth domain. The analysis of the fault plane solutions shows a complex stress field in the Vrancea crust, like a transition zone from the compressional regime at subcrustal depths to extensional regime characteristic

3 Probabilistic seismic hazard assessment in Romania 577 Fig. 1 Seismic crustal active zones in Romania and adjacent areas and their characteristics. for the entire Moesian platform [2]. The largest earthquakes, for which the fault plane solutions could be relatively well constrained, are the main shocks of the sequences of February 1983 (M w = 3.5), April 1986 (M w = 3.7), August September 1991 (M w = 3.5) and December 1997 (M w = 3.2) generated in the Râmnicu Sarat region [3]. 2.2. PREDOBROGEAN AND BARLAD DEPRESSION SOURCES Predobrogean Depression (PD) zone belongs to the southern margin of Predobrogean Depression marked by Sfantul Gheorghe fault (Fig. 1). Only moderate-size events are observed (M w 5.3) clustered especially along Sfantul Gheorghe fault. The fault plane solutions reflect the existence of the extensional regime of the deformation field. In our opinion this consistently reflects the affiliation of the Predobrogean Depression to the Scythian platform tectonic unit. The rate of the seismic moment release is Mo = 1.8 1015 Nm/year. The maximum observed magnitude for the Predobrogean Depression crustal zone is M w = 5.3, assigned to the event occurred on February 11, 1871. Barlad Deppression (BD), situated NE of the Vrancea zone, is characterized only by moderate size events (only four shocks with M w > 5.0, but not exceedind M w = 5.6). Considering that from seismotectonic point of view the Predobrogean

578 I. A. Moldovan, E. Popescu, A. Constantin 4 Depression belongs to the Scythian platform as well as Barlad Depression we considered the observed maximum magnitude for both zones, M w = 5.6. 2.3. INTRAMOESIAN FAULT The Intramoesian fault (IMF) crosses the Moesian platform in a SE-NW direction, separating two distinct sectors with different constitution and structure of the basement. Although it is a well-defined deep fault, reaching the base of the lithosphere [4], and extending southeast to the Anatolian fault region [5], the associated seismic activity is scarce and weak. Geological and geotectonic data indicate only a relatively small active sector in the Romanian Plain, situated to the NE from Bucharest. The geometry of the Intramoesian fault source and the distribution of the earthquakes with M w 3.0 occurred between 1892 and 2001 (30 events) are presented in Fig. 1. The magnitude domain of earthquakes is M w [3.0, 5.4]. The maximum magnitude was recorded in January 4, 1960 (M w = 5.4) in the central part of the Romanian Plain. 2.4. DULOVO AND SHABLA SOURCES A significant increase of seismicity is observed in the Dulovo and Shabla region (DU and SH), NE Bulgaria, where an earthquake with an estimated magnitude of M w = 7.2 occurred in 1901. The focal depth, whenever it can be constrained, has relatively large values (h ~ 35 km), suggesting active processes in the lower crust or in the upper lithosphere. Fig. 1 presents the geometry of the Intramoesian fault source and the distribution of the earthquakes with M w 3.0 occurred between 1892 and 2001 (20 events). The magnitude domain of earthquakes is M w [3.0, 7.2]. We pointed out that the greatest magnitudes are attributed to the two historical earthquakes: October, 14, 1892 (M w = 6.5) and March, 31, 1901 (M w = 7.2) and no event with magnitude greater than 5 was reported after 1950 since then the instrumental earthquake monitoring has become more used. 2.5. FAGARAS-CAMPULUNG-SINAIA AND TRANSILVANIAN DEPPRESSION CRUSTAL SOURCES The sources are located in the Southern Carpathians, Romania, adjacently to the West of Vrancea seismic region and are part of the major dome uplift of the Getic Domain basement. They are bordered at Northern and Southern edges by first order crustal fractures and consist of three seismogenic zones: Fagaras

5 Probabilistic seismic hazard assessment in Romania 579 zone containing Lovistea Depression and North Oltenia (FG), Campulung (CP) and Sinaia (SI) zones. The earthquake activity is related to intracrustal fractures extending from 5 to 30 km depth. The earthquakes in this zone are generated at South, on deep fractures extending on inherited hercynian lines along NW and NE alpine origin directions and at North, throughout Transylvania, along a stepped fault system separating the Carpathian orogen from its intermountain depression. In the western part of Fagaras Mountains, the earthquakes have a typical polikinetic character, with many delayed aftershocks, especially for large events, as the one produced in 1916. Preferential centers and lines of seismicity were identified after the occurrence of the large earthquakes and the subsequent aftershock activity. Most of the earthquakes are of low energy, but once per century a large destructive event with epicentral intensity larger than VIII is expected in Fagaras area. The last major shock occurred in January 26, 1916, M w = 6.5, Io = VIII IX. Fagaras seismogenic region is the second seismic source in Romania as concerns the largest observed magnitude (M w = 6.5), after the Vrancea intermediate-depth source (with maximum magnitude M w ~ 7.7 7.8). The Transilvanian Deppression (TD) seismogenic zone is defined only based on historical information, with the maximum reported earthquake M w = 6.5. The seismic activity at present is mostly absent. 2.7. THE CRUSTAL SOURCES FROM THE WESTERN PART OF ROMANIA The western and southwestern territory of Romania is the most important region of the country as concerns the seismic hazard determined by crustal earthquakes sources. The seismic risk in the region is also very high due to local risk factors and vulnerabilities: weak dwellings, old and unprotected buildings in the large cities, dams and chemical factories, high density of localities, great towns, and so on. The seismogenic Danubian zone (DA) represents the western extremity, adjacent to the Danube river, of the orogenic unit of the Southern Carpathians. The rate of seismic activity is relatively high, especially at the border and beyond the border with Serbia, across the Danube river. The magnitude does not exceed 5.6. The fault plane solutions are available for three earthquakes (the largest earthquake M w = 5.6 occurred in July 18, 1991) and indicate normal faulting with the T axis striking roughly N-S, in agreement with the general extensional stress regime in the Southern Carpathians. The contact between the Panonnian Depression and the Carpathian orogen lies entirely along the western part of the Romanian border. Even if no significant tectonic or geostructural differences are noticed, two enhancements in

580 I. A. Moldovan, E. Popescu, A. Constantin 6 the seismicity distribution can be identified in two relatively distinct active areas: Banat zone (BA) to the south, and Crisana-Maramures zone (CM) to the north. The seismicity of the Banat zone is characterized by many earthquakes with magnitude M w > 5, but not exceeding 5.6. The largest earthquake occurred after 1900 is the one from July 12, 1991 (M w = 5.6). Historical information suggests potential earthquakes greater than 6 in Crisana-Maramures, but only one event approaching magnitude 5 was reported in this century. The largest reported earthquake was M w = 6.5 on October 15, 1834. The Serbian seismogenic source named IBAR [6] is characterised by the occurrence of numerous crustal earthquakes with M w > 5.0. The largest earthquake occurred in the zone on April 08, 1893, has the magnitude Ms = 6.6 [7]. 3. THE FREQUENCY-MAGNITUDE DISTRIBUTION FOR THE DEFINED SOURCES The frequency-magnitude distribution for Vrancea crustal earthquakes is determined on the magnitude interval [3.0, 5.2]: lgn cum = (0.91 0.08)M w + (4.98 0.32) (1) with the correlation coefficient R = 0.98 and the standard deviation = 0.18. The distributions are plotted in Figs. 2 and 3. For Predobrogean Depression zone the frequency-magnitude distribution, estimated for the magnitude interval [3.0, 5.5], is presented in relation (2): lgn cum = (0.65 0.06)M w + (3.55 0.26) (2) with the correlation coefficient R = 0.97 and the standard deviation = 0.15. The distributions are plotted in Fig. 4. Fig. 2 (a) necumulative; (b) cumulative for Vrancea crustal earthquakes.

7 Probabilistic seismic hazard assessment in Romania 581 Fig. 3 Reccurence relation and magnitude distribution function for Vrancea earthquakes. Fig. 4 (a) necumulative; (b) cumulative relation for earthquakes occurred in Predobrogean Depression zone. The frequency-magnitude distribution for Barlad Depression zone is determined on the magnitude interval [2.5, 5.5]: logn cum = (0.75 0.05)M w + (4.15 0.20) (3) with the correlation coefficient R = 0.99 and the standard deviation = 0.1, and the annual number of earthquakes: 1.53479 eq/year. The distributions are plotted in Figs. 5 and 6. The frequency-magnitude distribution for Intramoesian fault-shabla- Dulovo crustal sources, determined for the magnitude interval [4.5, 7.2], is presented in relation (4) and plotted in Fig. 7: logn cum = (0.51 0.03)M w + (3.54 0.17) (4) with R = 0.99 and = 0.08.

582 I. A. Moldovan, E. Popescu, A. Constantin 8 Fig. 5 (a) necumulative; (b) cumulative relation for earthquakes occurred in Barlad Depression zone. Fig. 6 Reccurence relation and magnitude distribution function for earthquakes occurred in Barlad Depression zone. The frequency-magnitude distribution for Fagaras-Campulung-Sinaia crustal source regions for a magnitude interval [4.0, 6.5], leads to equation (5): lgn cum = (0.67 0.21)M w + (4.62 0.26) (5) with the correlation coefficient R = 0.91 and the standard deviation = 0.24, for M > 5.0, and lgn cum = (0.47 0.1)M w + (3.38 0.56) (6) with the correlation coefficient R = 0.91 and the standard deviation = 0.21, for M > 4.0. The distributions are both plotted in Fig. 8. The frequency-magnitude distribution for crustal earthquakes ocurred in Crisana-Maramures zone, was computed for two distinct regions CSM1 (equation 7, Table 1) and CSM2 (equation 8, Table 1) for a magnitude interval of [4.0, 5.6]:

9 Probabilistic seismic hazard assessment in Romania 583 lgn cum = ( 0,56 0.15)I + (4,6673 0.23) (7) with R = 0.91 and = 0.24 and lgn cum = ( 0,46 0.18)I + (3,670 0.25) (8) with R = 0.90 and = 0.21. Fig. 7 (a) necumulative; (b) cumulative (c) reccurence relation for earthquakes occurred in IMF-Shabla-Dulovo zone.

584 I. A. Moldovan, E. Popescu, A. Constantin 10 Fig. 8 The frequency-magnitude distribution for Fagaras-Campulung- Sinaia crustal sources, for 2 sets of data. Banat region on the magnitude interval [4.0, 5.6], for the entire time interval of the catalogue (equation 9) and after 1900 (equation 10): lgn cum = (0.82 0.08)M w + (4.73 0.38) (9) with the correlation coefficient R = 0.98 and the = 0.12. lgn cum = (0.74 0.06)M w + (4.31 0.27) (10) with the correlation coefficient R = 0.99 and the = 0.09. Danubian crustal earthquakes was determined for magnitudes between [4.0, 5.6] for two time intervals. One for the whole catalogue of earthquakes (equation 11) and the other for the earthquakes occurred after 1900: lgn cum = (0.82 0.09)M w + (4.94 0.46) (11) with R = 0.97 and = 0.15. lgn cum = (0.62 0.06)M w + (3.74 0.31) (12) with R = 0.98 and = 0.10. The necumulative and cumulative distributions are plotted in Fig. 9, for both time intervals and regions. The frequency-magnitude distribution for the Serbian seismogenic source named IBAR is estimated for the magnitude interval [3.7, 6.6] in equation (13) and for the intensity interval [4.5, 9.0] in equation (14) lgn cum = (0.87 0.03)Ms + (5.68 0.18) (13) with R = 0.99 and = 0.11, lgn cum = (0.50 0.03)Io + (4.82 0.19) (14) with R = 0.98 and = 0.18, and are both plotted in Fig. 10.

11 Probabilistic seismic hazard assessment in Romania 585 Fig. 9 necumulative; cumulative relation for Banat and Danuabian crustal sources ( M = 0.3). Fig. 10 The frequency-magnitude distribution for IBAR zone. In Table 1 are presented the characteristics of each source that will be used in the seismic hazard assessment. In the table the Fagaras-Campulung-Sinaia zone have been devided in 3 subzones: FG, CP and SI. The i value has been computed from the intensity-frequency distribution, using the formula from equation (15): i b ln10 (15)

586 I. A. Moldovan, E. Popescu, A. Constantin 12 Table 1 Input parameters for probabilistic hazard assessment using crustal sources Seismic Coordinates sources VRN 25.70/45.20-26.60/44.80 26.82/46.10-27.72/45.87 PD 27.45/45.85-29.00/44.90 27.72/45.87-29.30/45.20 BD 25.70/45.20-26.60/44.80 26.82/46.10-27.72/45.87 IMF 26.06/44.76-27.36/44.00 26.39/45.14-27.33/44.80 DUL 26.70/43.50-27.50/43.50 27.00/43.92-27.92/43.78 SH 28.03/43.36-28.73/43.28 28.31/43.72-28.90/43.71 FG 24.10/45.40-24.60/45.15 24.10/45.90-24.80/45.60 CP 24.95/45.00-25.30/45.10 24.95/45.50-25.30/45.50 TD 23.50/46.00-24.50/46.10 23.50/46.90-24.50/46.90 SI 25.05/45.70-25.50/45.40 25.20/45.80-25.50/45.60 Average depth Mmin Mmax b Imin Imax bi i Activity rate 20 3.0 5.9 0.91 2.5 6.5 0.60 1.38293 0.514526 10 3.0 5.6 0.65 2.5 6.5 0.42 0.99011 0.360254 10 2.5 5.6 0.75 2.0 6.5 0.49 1.12826 1.534712 15 4.5 5.4 0.51 5.0 6.5 0.32 0.74466 0.034600 15 4.5 7.2 0.51 5.0 9.0 0.32 0.74466 0.028000 15 4.5 7.2 0.51 5.0 9.0 0.32 0.74466 0.092600 15 4.0 6.5 0.76 5.0 8.5 0.50 1.15325 0.247403 15 4.0 5.0 0.66 5.0 6.0 0.44 1.01820 0.0865384 10 4.0 6.5 0.89 5.0 7.0 0.59 1.36774 0.010254 15 4.0 4.9 0.65 4.5 6.0 0.43 0.98781 0.076923 (continues)

13 Probabilistic seismic hazard assessment in Romania 587 Table 1 (continued) Seismic Coordinates sources CMS1 47.34/21.48-48.31/23.00 47.96/24.84-47.00/23.36 CMS2 48.26/19.60-48.72/22.72 48.24/23.24-47.81/20.24 DAN 21.00/44.90-22.30/44.15 21.80/45.70-23.00/44.80 BAN 21.00/44.90-21.90/45.80 20.70/4610-21.30/46.40 IBAR 19.80/44.00-20.80/44.60 21.00/43.10-21.80/43.80 Average depth Mmin Mmax b Imin Imax bi i Activity rate 15 4.0 6.5 0.56 4.5 8.0 0.56 1.29815 0.3762 15 4.0 6.5 0.46 5.0 8.0 0.46 1.06699 0.1027 15 4.0 5.6 0.71 5.5 9.0 0.43 0.98091 0.276700 10 4.0 5.6 0.82 5.5 9.0 0.50 1.16056 0.128800 14 3.7 6.6 0.87 4.5 9.0 0.54 1.24340 0.556203 4. ATTENUATION LAWS It is essential, for a probabilistic estimation of the seismic hazard, to constrain as much as possible how the energy of the seismic waves attenuates when propagating from the source to the site. The attenuation law for the crustal sources is given in the equation (16): I = Io c 1 log(d h /h) c 2 log(e) (D h h); (16) where: c 1, c 2 and a are different for each region; log(e) = 0.006514; D h is the hypocentral distance, and h is the depth presented, for each seismic source, in Table 1. For Vrancea, Barlad and Predobrogean Deppression we have used: c 1 = 3.16, c 2 = 3.02 and a = 0.0015 (1/m). For Fagaras-Campulung-Sinaia and Transilvanian Depression zones we have used: c 1 = 3.46, c 2 = 3.12 and a = 0.0013 (1/m). For the active zones from the southern and western part of Romania (IMF, BA, DA, CSM1 and CSM2), from Bulgaria and from Serbia (IBAR) we have used the attenuation law (equation 17) obtained by [8], with c 1 = 3.0, c 2 = 3.0 and = 0.0015 (1/m). I = I 0 3.0log(D h /h) 3.0 log(e) (D h h) (17)

588 I. A. Moldovan, E. Popescu, A. Constantin 14 5. CRUSTAL SEISMIC HAZARD ASSESSMENT For the input data set obtained in the present work, we applied the algorithm of [1] to compute the seismic hazard map of Romania in the case of crustal earthquakes. We present, in the following pictures, the hazard maps in terms of macroseismic intensities for different return periods (50, 100 years Fig. 11, 150, 200 and 475 years Fig. 12 and 1000 and 2000 years Fig. 13). The hazard values are in good agreement with the deterministic approach. If we compare our results for this case for 50 years return period with the computed intensities (converted from the peak ground displacement, velocity or acceleration) obtained by [9] using the deterministic approach, the differences are not exceeding 0.2 degrees in intensity, less than 0.5, the minimum measuring unit for intensities. We see that all the input parameters are directly related to the hazard pattern, as expected, and each of them is a crucial parameter in the seismic Fig. 11 Hazard map for return periods of 50 and 100 years.

Fig. 12 Seismic hazard curves for Romania, in terms of macroseismic intensities for return periods of 150, 200 and 475 years using only crustal seismic active zones.

590 I. A. Moldovan, E. Popescu, A. Constantin 16 Fig. 13 Seismic hazard curves for Romania, in terms of macroseismic intensities for return periods of 1000 and 2000 years using only crustal seismic active zones. hazard probabilistic approach. The Romanian seismic data used in the paper were obtained from [10] and outside the borders from [7] and [11]. 6. CONCLUSIONS This work is a useful tool for the assessment of the seismic risk and implementation of antiseismic protection measures in the case of special constructions and strategic objectives, such as, nuclear power plants (Cernavoda NPP), large cities situated outside the Vrancea intermediate seismic source influence zone (Timisoara, Turnu Severin, Arad, Cluj) and hidroenergetic large constructions.

17 Probabilistic seismic hazard assessment in Romania 591 REFERENCES 1. R. K. McGuire, EQRISK-Evaluation of Earthquake Risk to Sites, United States Department of the Interior, USGS, Open File Report No. 76 67, 1976. 2. M. Radulian, N. Mândrescu, G. F. Panza, E. Popescu, A. Utale, Characterization of seismogenic zones of Romania, Pure Appl. Geophys., 157, 57 77, 2000. 3. E. Popescu, M. Radulian, Source characteristics of the seismic sequences in the Eastern Carpathians foredeep region (Romania), Tectonophysics, 338, 325 337, 2001. 4. D. Enescu, B. D. Enescu, Contributions to the knowledge of the genesis of Vrancea Earthquakes, Romanian Reports in Phys. 45, 777 796, 1993. 5. M. Sandulescu, Geotectonics of Romania. Ed. Tehnica, Bucuresti, 1984, p. 334 (in Romanian). 6. R. M. W. Musson, Generalised seismic hazard maps for the Pannonian Basin using probabilistic methods, Pure Appl. Geophys., 157, 147 169, 2000. 7. N. V. Shebalin, G. Leydecker, N. G. Mokrushina, R. E. Tatevossian, O. O. Erteleva, V. Yu. Vassiliev, Earthquake Catalogue for Central and Southeastern Europe 342 BC - 1990 AD. European Commission, Report No. ETNU CT 93-0087, Brussels, 1998. 8. T. Zsiros, Macroseismic focal depth and intensity attenuation in the Carpathian Region, Acta Geod. Geophys. Hung., 31 (1 2), 115 125, 1996. 9. M. Radulian, F. Vaccari, N. Mandrescu, G. F. Panza, C. Moldoveanu, Seismic Hazard of the Circum-Pannonian Region, eds. G. F. Panza, M. Radulian, C.-I. Trifu, Pure Appl. Geophys. 157, 221 247, 2000. 10. ***, Romplus catalog-updated until 2006 by the department of Data Acquisition of the National Institute for Earth Physics-Bucharest. 11. ***, International Seismological Center, On-Line Bulletin, http://www.isc.ac.uk, Thatcham, U.K.