SEISMICITY AND SEISMOLOGICAL OBSERVATIONS OF EXCURSION GUIDE AND ADJACENT TERRITORIES

International Workshop

SS S EEEIIISSSMMMIIICCCIIITTTYYY AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION AND ADJACENT TERRITORIES

September 10–12, 2007 Vilnius,

VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS

Vilnius, 2007 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

International Workshop “Seismicity and seismological observations of the Baltic Sea region and adjacent territories”, September 10–12, 2007, Vilnius, Lithuania: Volume of abstracts / Compiled by. J. Lazauskienė, J. Satkūnas; International Union of Geological Sciences (IUGS), International Borders – Geoenvironmental Concerns (IBC), Lithuanian Geological Survey. – Vilnius: LGT, 2007. – 80 p.: iliustr.

Workshop is held under auspice of IUGS–GEM working group – International Borders – Geoenvironmental Concerns (IBC)

Organising Committee Chairman: Dr. Jonas Satkūnas, Lithuanian Geological Survey Secretary: Dr. Jurga Lazauskienė, Lithuanian Geological Survey Members: Dr. Jolanta Čyžienė, Lithuanian Geological Survey M. Sc. Andrius Pačėsa, Lithuanian Geological Survey

Advisory board Valērijs Ņikuļins, Latvian Environmental, Geological and Meteorological Agency, Riga, Prof. Dr. Habil. Saulius Šliaupa, Vilnius University, Vilnius, Lithuania

Structure and aims of the workshop The workshop is devoted to the seismicity and seismological observations of the Baltic Sea region and adjacent territories: historical and present situation, installation of new seismic stations, processing and analysis of seismic data, seismic hazard assessment and risk mitigation. The major aim of the workshop is to disseminate and discuss the major seismicity related issues over the whole Baltic Sea region and the wide range of issues related to the low seismicity , focusing on networking, monitoring, assessment of seismicity, data management and scientific co‐operation.

Compiled by: J. Lazauskienė, J. Satkūnas Layout: R. Norvaišienė Cover design: I. Virbickienė

Circulation: 50 copies

© Lietuvos geologijos tarnyba, 2007

222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

CONTENT

DISCREPANCY IN LOCATION OF MACROSEISMICALLY AND INSTRUMENTALLY DERIVED EPICENTERS OF , SEPTEMBER 21, 2004, EARTHQUAKE – FACTUAL DATA AND POSSIBLE EXPLANATIONS...... 5 Aleshin A. S., Aptikaev F. F., Nikonov A. A., Pogrebchenko V. V. SEISMOLOGICAL OBSERVATIONS IN ESTONIA ...... 8 All T. MODERN MICROSEISMIC OBSERVATIONS IN KALININGRAD REGION (TRACKING CONSEQUENCES OF 2004 EARTHQUAK) ...... 10 Ambrosimov A.K., Kovachev S.A., Sivkov V.V. SEISMICITY AND SEISMIC OBSERVATIONS IN BELARUS ...... 11 Aronov A.G., Aronova T.I., Kulich O.N. KALININGRAD EARTHQUAKE SEPTEMBER 21 2004 – TECTONIC MODEL...... 14 Assinovskaya B., Ovsov M., Zhamoida V., Shcherbakova N. THE NEW SWEDISH NATIONAL SEISMOLOGICAL NETWORK ...... 15 Bödvarsson R. NEW PERSPECTIVES OF MONITORING OF THE SEISMICITY OF GULF OF GDANSK AND ADJACENT AREAS...... 16 Dębski W., Wiejacz P., Suchcicki J., Wiszniowski J. SOURCE PARAMETERS OF KALININGRAD EARTHQUAKES ...... 17 Domański B., Dębski W. RECONSTRUCTION OF STRESS FIELDS IN ADJACENT REGIONS SEPARATED BY SEISMICALLY-ACTIVE FAULTS ...... 18 Galybin A. N., Mukhamediev Sh. A. CORRELATIONS OF MAGNITUDE AND FELT-AREA FOR EARTHQUAKES IN THE FENNOSCANDIAN SHIELD/EAST EUROPEAN PLATFORM ...... 22 Gregersen S., Husebye E., Mantyniemi P. NORWEGIAN NATIONAL SEISMIC NETWORK RECORDING EVENTS IN THE BALTIC COUNTRIES...... 24 Havskov J. LITHUANIAN SEISMIC NETWORK – CURRENT STATUS AND PERSPECTIVES ...... 28 Lazauskienė J. SELENA; A SOFTWARE FOR NEAR REAL-TIME DAMAGE ESTIMATION AND DAMAGE SCENARIOS ...... 31 Lindholm, C. D., Molina S., Bungum H., Oye V. A 2-D SEISMIC SIGNAL DETECTOR FOR STAND ALONE 3-COMPONENT STATIONS ...... 38 Matveeva T., Fedorenko Yu.V., Husebye E.S. THE KARELIAN REGIONAL SEISMIC NETWORK – THE COSSACK RANGER II SEISMOGRAPH 43 Matveeva T., Fedorenko YU. V., Husebye E. S. AUTOMATIC P-CODA PHASES IDENTIFICATION USING BAYESIAN APPROACH...... 47 Matveeva T., Fedorenko Yu. V., Fedorenko M., Husebye E. S. SEISMIC RESISTANCE OF SYSTEMS AND ELEMENTS OF EXISTING IGNALINA NPP ON THE BASIS OF PERFORMED INVESTIGATIONS ...... 51 Mereznikov A.

333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SEISMIC HAZARD OF TERRITORY OF LOCATION OF THE IGNALINA NUCLEAR POWER PLANT BY RESULTS OF COMPLEX RESEARCHES 1987–1988 OF THE LAST CENTURY... 53 Mindel I.G., Trifonov B.A. APPROACH TO PARAMETRIZATION OF TECTONIC EARTHQUAKES WITHIN THE KALININGRAD DISTRICT, , BY MACROSEISMIC DATA ...... 57 Nikonov A. A. KALININGRAD, SEPTEMBER 21, 2004, EARTHQUAKE IN THE EASTERN BALTIC AREA – BASIC MACROSEISMIC MAPS FOR THREE MAIN SHOCKS ...... 60 Nikonov A.A., Pačėsa A., Aptikaev F.F., Nikulin V.G., Puura V., Aronov A.G. REGIONAL FEATURES OF SEISMOTECTONICS AND DEFORMATION OF EARTH CRUST OF ...... 63 Nikulin V. G. SEISMOLOGICAL OBSERVATIONS IN LITHUANIA...... 66 Pačėsa A. APPLICATION OF THE PROBABILISTIC APPROACH IN ASSESSMENT OF THE SEISMIC HAZARD OF THE BALTIC REGION ...... 69 Pačėsa A., Šliaupa S. INSTRUMENTATION OF THE SEISMIC ALARM AND MONITORING SYSTEMS OF IGNALINA NUCLEAR POWER PLANT ...... 71 Razinkov O., Epp M., Kündig C., Davidiuk O., Narbuntas J. SEISMIC MONITORING OF AN UNDERGROUND NUCLEAR WASTE REPOSITORY AT OLKILUOTO, FINLAND ...... 74 Saari J. THE ROUTINE MICROEARTHQUAKE ANALYSIS PACKAGE IMPLEMENTED IN ICELAND (IMO) SND SWEDEN (SNSN, UU)...... 76 Slunga R. REASSESSMENT OF THE DESIGN BASIS EARTHQUAKE FOR IGNALINA NPP, NE LITHUANIA 77 Šliaupa S., Kačianauskas R., Markauskas D., Dundulis G. NORWEGIAN NATIONAL SEISMIC NETWORK REAL TIME MONITORING...... 79 Utheim T., Havskov J. THE SEISMOLOGICAL NETWORK IN DENMARK AND IN GREENLAND, EARTHQUAKE MONITORING AND APPLIED RESEARCH ...... 80 Voss P.

444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

DDDIIISSSCCCRRREEEPPPAAANNNCCCYYY IIINNN LLLOOOCCCAAATTTIIIOOONNN OOOFFF MMMAAACCCRRROOOSSSEEIIISSSMMMIIICCCAAALLLLLLYYY AAANNNDDD IIINNNSSSTTTRRRUUUMMMEEENNNTTTAAALLLLLLYYY DDDEEERRRIIIVVVEEEDDD EEEPPPIIICCCEEENNNTTTEEERRRSSS OOOFFF KKKAAALLLIIINNNIIINNNGGGRRRAAADDD,,, SSSEEEPPPTTTEEEMMMBBBEEERRR 222111,,, 222000000444,,, EEEAAARRRTTTHHHQQQUUUAAAKKKEEE ––– FFFAAACCCTTUUUAAALLL DDDAAATTTAAA AAANNNDDD PPPOOOSSSSSSIIIBBBLLLEEE EEEXXXPPPLLLAAANNNAAATTTIIIOOONNNSSS

Aleshin A. S., Aptikaev F. F., Nikonov A. A., Pogrebchenko V. V. IFZ RAS, Bol. Gruzinskaya, 10, Moscow, Russia; [email protected]

A great number of articles [1–4 and others] are dedicated to examination of peculiarities of Kaliningrad earthquakes on September 21, 2004. At the same time some problems required more clarity. The most important of them to our opinion is discrepancy in focal locations by macroseismic and instrumental data. This problem is important from the point of view of connection of tectonic structures with the focal zones of recent earthquake and also the determination of potential earthquakes zones. Examination of these discrepancies and their possible causes are the subject of this report. Macroseismic evaluations Macroseismic evaluations of focal zones of each of main shocks of Kaliningrad earthquakes 21.09.2004 may be made by the following collection maps analysis: – maps of intensity and isoseismals; – maps of sensitive oscillations directions; maps of directions and power sound; – maps of inclines, objects and constructions fall directions. Conclusion about three main shocks epicenters locations is based on the following facts [1]. Maximal shaking zones of three main shocks are limited towards the land on Sambia Peninsula but open toward the sea. Maximal isoseismals in each case are limited so that its main area and consequent geometrical centers are in adjacent aquatoria. Regular incrimination of intensity shaking toward interior of Sambia Peninsula and general correspondence of minor and maximal isoseismals make impossible localization of epicentral zones in Sambia Peninsula interior. The maximal shaking zones stretching out are in meridian directions along Sambia Peninsula west coast by first and third shocks and sublatitude directions along Sambia Peninsula north coast by second shock are interpreted by well known regularities as consequence focal zones orientations. Maximal isoseismal contours for each of the three shocks determine first possible limits of localization epicenters and second determine epicenters localizations approximately some km from coast toward the sea. Geological data are important independent from macroseismic observations indication that Kaliningrad earthquakes 21 09 2004 focal zones are under sea bottom along the Sambia Peninsula. At west and north Sambia Peninsula massif limits are first of all remarkable morphological ledges by some tens meters height. No less important is that similar paralleled to the coast ledges are at adjacent bottom parts. This together with the fact of different heights of the tops of Paleogene and Neogene deposits of the land and the adjacent bottom show existence of modern faults with no less 10 meters amplitude in coastal area. Such facts were found by marine seismoacoustical sounding along north and west coasts of Sambia Peninsula in 2–3 km

555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

from present coastal line. It is very probable that these faults were made active by Kaliningrad earthquakes 21 09 2004. On the contrary in the Sambia Peninsula limits by relief or geological features of faults zones those activating could be dated to the time of recent glacial epoch or later that is younger 25–20 thousand years do not exist. In general there is a number of macroseismic indicators about location of focal zones of Kaliningrad earthquakes 21 09 2004 in sea along north and west coasts of Sambia Peninsula. Instrumental evaluations Locations of three main shocks are evaluated by seismic wave records by several seismological services: ƒ European – Mediterranean Centre – EMSC; ƒ National Earthquake Information Center of States Geological Survey – NEIC; ƒ Geophysical Survey of Russian Academy of Science – GSRAS; ƒ Institute of Geophysics of – IGF. Both instrumental and macroseismic evaluations results are showed on the Figure.

Figure. The instrumental and macroseismic evaluations of locations of Kaliningrad earthquakes on September 21, 2004 epicenters.

First that should be remarked is the considerable scattering in particular by the first shock. Generally instrumental evaluation scattering is limited by the rectangle with sides up to

666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

40 km by longitude and 30 km by latitude which is remarkably more than precision given by corresponding surveys. Each of services may possibly use mean square data derivation as means of precision instrument not including systematic derivation. Partly this phenomenon is explained by the absence near epicentral zones of seismic stations. The nearest Suwalki station (Poland) is in 220 km distance from epicentral zone. Utilization of different time – distance curves IASPEI 91 and AK 135 gives remarkable difference in location even of the same shock. By this cause we consider unsuccessful that determination of epicenter location by probabilistic approach as that was made in article [2]. This probabilistic evaluation of epicenter location was accepted as general and figures in consequent works [3, 4] show more conformed picture. According to instrumental evaluations epicenters locations of Kaliningrad earthquakes 21.09.2004 are in the center of west part of Sambia Peninsula. This part of Sambia Peninsula has no macroseismic features which are in accordance with instrumental evaluation of epicenter location. Moreover there are no geological (seismotectonic) structures that may be taken into account as active and coordinate with discussed earthquakes. The reason of these sharp differences in Kaliningrad earthquakes 21 09 2004 epicenters locations in our opinion is remarkable velocity inhomogeneity of this region which reveals itself as consequence of geological and geomorphologic difference of structure (on north‐ west – sea; on south‐east – land) and complex stress‐strain conditions. These circumstances cause important mistakes by using some whatever travel time model. Velocity models of this region, apparently, are different for different directions. In any case, the noted differences require further deeper investigations. REFERENCES:

1. Nikonov A.A., Aptikaev F.F., Aleshin A.S., Pogrebchenko V.V., Erteleva O.O., Assinovskaya B.A. The investigation of the Kaliningrad, September 21, 2004, earthquake consequences (in Russian) // http: //www.scgis.ru/russian/cp1251/h_dgggms/1‐2005/screp‐3.pdf 2. Wiejacz P. 2004. Preliminary investigation of the September 21, 2004 earthquakes of Kaliningrad Region, Russia, Acta Geophys. Pol. 52, 425‐441 3. Маловичко А.А., Мехрюшев Д.Ю., Старовойт О.Е., Габсатарова И.П., Чепкунас Л.С. О Калининградских землетрясениях 21 сентября 2004 гю и развитии сейсмического мониторинга в Калининградской области // Современные методы обработки и интерпретации сейсмологических данных, Обнинск, 2006, стр. 88–97 4. Gregersen, S., Mäntyniemi, P., Nikonov, A.A., Aptikaev, F.F., Aleshin, A.S., Assinovskaya, B.A., Pogrebchenko, V.V., Guterch, B., Nikulin, V., Pacesa, A., Wahlström, R., Schweitzer, J., Kulhánek, O., Holmquist, C., Heinloo, O., Puura, V., 2005, Felt reports at large distances of teh earthquakes in non‐seismic Kaliningrad in West Russia, In: The Kaliningrad earthquake September 21, 2004 workshop materials, A. Jõeleht (ed.), University of Tartu, 11–12

777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMOOOLLLOOOGGGIIICCCAAALLL OOOBBBSSSEEERRRVVVAAATTTIIIOOONNNSSS IIINNN EEESSSTTTOONNNIIIAAA

All T. Geological Survey of Estonia, Kadaka str. 82, 12618 Tallinn, ESTONIA e‐mail: [email protected]

Introduction In Estonia the Seismic Monitoring Program forms a subprogram of the National Environmental Monitoring Program financed by the Ministry of the Environment, since 1996. The monitoring network is in the possession of the Geological Survey of Estonia (GSE) since 1994. The Estonian National Seismic Monitoring Network presently consists of three stations: Vasula, Suurupi and Matsalu. The Vasula station belongs also to the GEOFON network. Network description Table 1 shows basic parameters of seismic stations of the National Network of Seismic Stations – status in August 2007. The locations of the stations are given in Figure 1. The Vasula Seismological Station belongs to the GEOFON network (http://www.gfz‐ potsdam.de/geofon/). Potsdam Geoinvestigation Center (GeoForschungsZentrum, Potsdam, GFZ) installed a Quanterra Q380 datalogger and STS‐2 seismometer in Tartu Old Astronomy Observatory in June 1996. The Seismological Station was included to the GEOFON‐project (GeoForschungsNetz) led by GFZ. International code of Tartu Seismological Station was TRTE. Because of intensive background, disturbing the investigations it was necessary to remove it outside of Tartu. The new location was selected to be in the nearby village Vasula (see Fig. 1). For now the observations are stopped at TRTE. On April 26, 2003 the STS‐2 seismometer with Earth Data digitizer and SeisComP PC was installed at Vasula. The Vasula station is registered under international code VSU and it belongs also to the GEOFON network. The station is equipped with RDSL internet connection. The data are transformed in real time to the EGS and GEOFON DC. The Suurupi Seismological Station on Suurupi Peninsula (Fig. 1) with short‐period transducers of SM‐3 type was initially oriented to detect only weak local seismic events. On August 19, 2003 the station was hit by lightning and was out of order until spring 2005. In April 2005 the Suurupi station was renovated and upgraded to the BB station. The station is now equipped with Güralp CMG‐6T transducer and Quanterra Q380 data logger. RDSL internet connection was also established to transform the continuous WD in miniSEED format to the EGS DC and GEOFON DC. In August 2006 the third monitoring station was installed in western Estonia at the center of Matsalu National Park. The new station is named Matsalu and was registered under the international code MTSE. The station is equipped with Güralp CMG‐6T transducer, Earth Data digitizer and SeisComP PC and RDSL internet connection.

888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 1. The location of Estonian seismological stations (VSU – Vasula seismological station, SRPE – Suurupi seismological station, MTSE – Matsalu seismological station)

Table 1. Basic parameters of existing seismic stations of the Estonian National Network of Seismic Stations

Name Code Latitude Longitude Seismometer DAS Sampling freq. Data transfer Data format Vasula VSU 58°27'60'' 26°44’40’’ STS-2 SeisComP 100 Continuous, mSEED real time Suurupi SRPE 59°27'48'' 24°22'48'' CMG-6T SeisComP 100 Continuous, mSEED real time Matsalu MTSE CMG-6T SeisComP 100 Continius, mSEED real time

Data storage, archiving and availability In all stations the ComServ performs the data acquisition and recording. The SeedLink server is used to transfer the data in mSEED format to the GSE and GEOFON Data Centers. The data are archived in both locations. The data about earthquakes are recorded also at IRIS (Incorporated Research Institutions for Seismology) DataCenter and are available from IRIS Data Sources. The online archive in GSE was established in December 2004. The Waveform data are available using the AutoDRM request. The AutoDRM address is [email protected] and a typical request‐mail looks like following: BEGIN GSE2.0 TIME 2005/03/16 16:00:00 TO 2005/03/16 16:10:00 STA_LIST TRTE,SRPE,VSU CHAN_LIST * WAVEFORM SEED STATION GSE2.0 CHANNEL GSE2.0 OUTAGE GSE2.0 EMAIL [email protected] STOP The analysis of the seismic data is performed by Seismic Handler package. The results of seismogram interpretation and earthquake location are stored in databases and published in annual bulletins.

999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

MMMOOODDDEEERRRNNN MMMIIICCCRRROOOSSSEEEIIISSSMMMIIICCC OOOBBBSSSEEERRRVVVAAATTTIIIOOONNNSSS IINNN KKKAAALLLIIINNNIIINNNGGGRRRAAADDD RRREEEGGGIIIOOONNN (((TTTRRRAAACCCKKKIIINNNGGG CCCOOONNNSSSEEEQQQUUUEEENNNCCCEEESSS OOOFFF 222000000444 EEEAAARRRTTTHHHQQQUUUAAAKKK)))

Ambrosimov A.K., Kovachev S.A., Sivkov V.V. P. P. Shirshov Institute of Oceanology Russisan Akademy of Sciences, [email protected]

At the end of November 2006 P.P. Shirshov Institute of Oceanology installed self‐contained bottom seismic station (OBS) at the bottom of Gulf of Danzig in Baltic Sea. Present researches were undertaken for studying background seismic activity in the western part of the Kaliningrad Region. In accordance with maps of general seismic zonation of russian territory (OCP‐97) the Kaliningrad Region is considered as a seismic region. However in September 2004 the series of seismic events was taken place here with magnitudes about 5 and has foci, which were located near the coastline of Gulf of Danzig. Jarring of the main shock in epicenter of the earthquake crested about 7 on the rotten ground and crested about 5 in some districts of Kaliningrad. The placing of bottom seismograph was the first stage in studying local zone of a real earthquake. Bottom seismograph is a self‐contained device, which has three geophones transducers (one – vertical and two – horizontal), hydrophone receiver and high‐cycle quartz clock. The registration of the information in this device is continually conducted on the flash‐card in digital format. The seismograph was installed up in 10 km from the coastline and at depth in 27 m. Registration of seismic events was taken in a period from December 2006 till February 2007. The previous data evaluation from this device let to locate series of teleseismic events with epicenter in Poland with magnitudes mb from 3.3 to 4.3. This series were outlaying from the device more, then 300 km. For another thing, bottom seismograph for 3 months of it’s working checked in about 20 very low seismic events with magnitudes (ML) from 0.3 to 1.5, which, judging by differences in time of advent shear and longitudinal waves (S‐P), were taken place on distance nearly 20–30 km from the device. These low seismic events could be considered, as micro earthquakes from the local zone of the strong earthquake in 2004 in the Kaliningrad Region. In May 2007 bottom‐dwelling seismograph was picked up, recharged and located back for the continuance the working. At the same time were located three land seismic stations. So, such network from scratch seismic stations in the Kaliningrad Region, consists of 4 registration points, which surrounded epicenter of sensible earthquake in 2004, what in future allows to determine positions of epicenters and the depths of the micro earthquake’s focuses, which confined to a zone of the powerfull seismic event in 2004.

111000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMIIICCCIIITTTYYY AAANNNDDD SSSEEEIIISSSMMMIIICCC OOOBBBSSSEEERRRVVVAAATTTIIIOOONNSSS IIINNN BBBEEELLLAAARRRUUUSSS

Aronov A.G., Aronova T.I., Kulich O.N. The Centre of Geophysical Monitoring of the National Academy of Sciences of Belarus, Kuprevich str, 7, Minsk, 220141, Republic of Belarus, e‐mail: [email protected]

The territory of Belarus is situated in the west of the old East European Platform (EEP) which involves the Baltic and Ukrainian shields, Russian and Volyn‐Azov plates and is rated as a low‐magnitude seismic zone according to the seismotectonic zoning [1, 2]. Belarus and the Baltic States comprise the single seismotectonic region described by the similar geological evolution and common recent geodynamic conditions. The region shows a rather low seismic activity, however some seismic events with a magnitude M ≤ 4.5 were recorded within its limits. Instrumental seismic observations started in the territory of Belarus in 1965 at the Pleshchenitsi geophysical observatory. At a later time the seismic network development was associated with investigations of the seismic regime and the territory division into the seismic regions, the study of deep structure.

Fig. 1. Map of epicenters of seismic events in the territory of Belarus within 1887–2007 (1 – earthquake magnitude; 2 – capital of Belarus; 3 – seismic station; 4 – town; 5 – frontier; 6 – date of earthquake). Seismological monitoring within the studied period was carried out by continuous observations at the sites as follow: geophysical observatories “Minsk” (Pleshchenitsi), “Naroch” and seismic stations “Soligorsk”(local network), “Brest”, “Gomel”, “Glushkevichi”, “Mogilev”, “Polotsk” (local network) [3.5]. The seismicity of the territory of Belarus has recently received a thorough study. Full advantages were taken of the results of continuous instrumental observations presented in bulletins of seismic stations within 1965–2007. When the data available in seismological bulletins of the seismic stations of Belarus were analyzed and summarized, a Catalogue of Earthquakes of the territory of Belarus since 1887 till 2007 was compiled with a due regard for

111111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

historical earthquakes revealed. A total of 1290 seismic events were instrumentally recorded and processed within the studied territory, for 1020 of them the epicenters were located by coordinates. The Catalogue comprises 1024 seismic events with the magnitude M ≤ 4.5 among them four historical earthquakes in 1887, 1893, 1896, 1908 [3, 4, 6, 7]. The Catalogue of Earthquakes of Belarus of 1887–2007 was used the basis for the Map of Epicenters of seismic events in the territory of Belarus (see Fig. 1). The size of circles in the map corresponds to the earthquake magnitude values from 0.6 to 4.5. An analysis of the position of epicenters shows their uneven distribution in area [5, 8]. In the northern part of the territory of Belarus only single shocks were recorded which is, probably, due to the impossibility to determine the position of many minor earthquakes because of a poor resolution of seismic observations, on the one hand, and a low seismic activity of this territory, on the another hand. At the same time felt historical earthquakes of 1887, 1893, 1896, 1908 occurred just in the northern part of the studied region. On December 10, 1887 the magnitude – 3.7 took place struck the region of Borisov, the intensity in the epicenter was as high as 6. In some places of the Borisov district (Minsk province) underground roaring similar to thundering was heard. Window‐panes were broken in many houses. For the first time after the publication in the XIX century the present paper describes two historicfl earthquakes that occurred near Mogilev. The paper by I.V. Mushketov [9] presents eye‐witness accouts of the about events. So, the first earthquake took place on 29 August, 1893 and measured 3.5 in magnitude and 5 in intensity and was described as follows: the first shock was shot, but strong, the secondary – not so strong, but more extended from northwest and lasted for no more than 6 seconds. An inspector of a school wrote in his report: the earthquake lasted for several seconds, sitting people felt chair shaking, heard clatter of crockery and slight roaring oil lamps on the table were shaking. To calculate the magnitudes of these events an equation of the macroseismic field relating the earthquake intensity in the epicenter, magnitude MLH and focal depth was used. On 28 December, 1908 the earthquake measuring 4.5 in magnitude took place at the settlement of Gudogai (Ostrovets district) and was felt in many nearby settlements. The shaking intensity at the farmstead of Serzhanti was as high as 7: violent roaring like thundering window‐panes jingling, land shaking, animals tumbling down and impression of falling house walls. A deep ditch running from the north to the southeast for a verst (3 500 feet) was formed there. The intensity of shaking at the village of Bystritsa was 5– 6: noise like a creaking loaded cart, repeated land shakes, house shaking as if it is moved, people awaked, a hand‐saw dropped down in some places land apectures. The magnitude – 2.5 earthquake occurred 2.8 km east of the settlement of Oreshkovichi (Velikoye Boloto stow) on 8 July, 1980. The earthquake of magnitude – 2.5 took place 10 km east of the town of Ostrovets (Ostrovets district) near the forest belt and the Losha River on 27 February, 1987. The earthquake of magnitude – 2.1 struck the forest belt 2 km southeast of the settlement of Kokhanovichi (Verkhnedvinsk district). Earthquake of magnitude of 1.1 and 0.8 occurred 2.8– 3.4 km east of the Zhartsi settlement (bog‐forest massif) in the Polotsk district. The magnitude of 1.3 earthquake took place southeast of the Selniki settlement (forest belt) in the Polotsk district on 8 November, 2006. Earthquake foci are numerous in the southern part of the territory of Belarus [8]. All these are confined to a zone where the northwestern part of the Pripyat Trough and Belorussian Anteclise are going. Areas of rather frequent minor seismicity cases run also northwards of this zone. The magnitude – 3.0 earthquake took place in the region of the Kulaki village near Soligorsk on 10 May, 1978. The earthquake intensity in the epicenter was 4‐5. Roaring and window‐pane jingling were heard, hanging things were swinging, furniture and

111222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

floor on the ground floors of wooden buildings were squeaking. The drift roof of a mine of the potassium group of mines fell down. The earthquake of magnitude – 2.8 occurred on 1 December, 1983 3 km south of the settlement of Povstan and 40 km northeast of the town of Soligorsk. The intensity of shaking in the earthquake epicenter was 4‐5. Roaring and window‐ pane jingling were heard, hanging things were swinging, floor and furniture were squeaking, fractures in building walls formed. The earthquake took place west of Glusk and 70–80 km east of Soligorsk on 17 October, 1985. The land was shaking, hanging things swinging, flowers shaking. The earthquake of magnitude – 1.9 occurred in the region of the Pogost village, Soligorsk district on 16 March, 1998. It was felt with the intensity of 4–5. The inhabitant of this region heard window‐pane jingling, floor squeaking, sleeping people awake, furniture was moving, doors opened spontaneously. These were fractures in the walls, somewhere Dutch tile dropped down. The shaking area was 1.5–2 km in radius. These factors are indicative of a small depth of the event. It is supposed that this seismic event could cause a spontaneous lava collapse, which was responsible together with the main event for the greatest impart. A day later on 17 March, 1998 the magnitude – 0.8 event took place there and was accompanied by similar, but less pronounced sensations. There were no more collapses in the mine. The earthquake of magnitude 2.5 occurred 2 km northwest of the settlement of Berezhtsi (forest track), Zhitkovichi district on 16 May, 1999, and the magnitude 2.8 earthquake was recorded on 13 May, 2005 3 km west of the settlement of Sergeyevichi (forest‐bog massif), Maryina Gorka district.

REFERENCES:

Aizberg R.Ie., Aronov A.G., Garetsky R.G., Karabanov A.K., Safronov O.N. Lithosphere, Minsk, 1997, № 7, 5–17 ( in Russian). Aizberg R.Ie., Garetsky R., Aronov A., Karabanov A.K., Safronov O.N. Technika roszukiwan geologisznych. Geosynoptika i geotermia. Krakow, 1999, XXXVIII ‐1(195), 28–37. Aronov A.G., Seroglasov R.R., Aronova T.I. Belarus // Earthquakes in Northern in 1997. Collection of Articles. Edit. Board: O.Ye. Starovoit (exec. edit) e al. Obninsk, 2003, 172‐180 (in Russian). Aronov A.G., Seroglasov R.R., Aronova T.I. Belarus // Earthquakes in Northern Eurasia in 1998. Collection of Articles. Edit. Board: O.Ye. Starovoit (exec. edit) e al. Obninsk, 2004, 188–194 (in Russian). Aronov A.G., Seroglasov R.R., Aronova T.I. Seismicity of the territory of Belarus. In: Earthquakes and microseismicity among challenges of recent geodynamic of the East European Platform. Edited by N/V/ Sharov, A.A. Malovichko, Yu.K. Shukin. Book 1. Earthquakes. Petrozavodsk. Karelian Scientific Centre, RAS, 2007, 357–364 (in Russian). Aronov A.G., Seroglasov R.R., Aronova T.I. Belarus // Earthquakes in Northern Eurasia in 1999. Collection of Articles. Edit. Board: O.Ye. Starovoit (exec. edit) e al. Obninsk, 2005, 200–203 (in Russian). Aronova T.I. Belarus // Earthquakes in Northern Eurasia in 2000. Collection of Articles. Edit. Board: O.Ye. Starovoit 9exec. Edit) e al. Obninsk, 2006, 199–204 (in Russian). Aronova T.I. Peculiar manifestations of seismotectonic processes in the territory of Belarus. Minsk, Lithosphere, 2006, № 2(25), 103–110 (in Russian). Mushketov I.V. Materials for the study of earthquakes in Russia. Attachment to the vol. XXXV of Transaction of the Imperial Russian Geographical Society. St.‐Petersburg, 1899, 91–102 (in Russian).

111333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

KKKAAALLLIIINNNIIINNNGGGRRRAAADDD EEEAAARRRTTTHHHQQQUUUAAAKKKEEE SSSEEEPPPTTTEEEMMMBBBEEERRR 22111 222000000444 ––– TTTEEECCCTTTOOONNNIIICCC MMMOOODDDEEELLL

Assinovskaya B1., Ovsov M2., Zhamoida V3., Shcherbakova N4. 1GAO RAS, Pulkovskoye sh., 65/1, Saint‐Petersburg, Russia, [email protected] 2GAO RAS, Pulkovskoye sh., 65/1, Saint‐Petersburg, Russia, [email protected] 3VSEGEI Sredni pr. 76, Saint‐Petersburg, Russia, [email protected] 4GAO RAS, Pulkovskoye sh., 65/1, Saint‐Petersburg, Russia, [email protected]

The Kaliningrad event source parameters are rather indefinably due to known objective reasons. In this connection, we tried to find new data to confirm earthquake source position proposed earlier. First, we have analyzed additional intensity evidences to compile alternative EMS intensity map of both Kaliningrad events source zones. Second, we collected available data of some moderate earthquakes occurred in European countries and USA where well constructed local seismic networks operate. The relationships between intensity fields of these events and different type instrumental positions were established in comparison with Kaliningrad earthquake parameters. Third, we researched all published regional and local geological information concerning tectonics, present‐day activity data, deep structure, potentional fields of the north‐western part of the Sambian Peninisula. Two kinds of modelling based on gravity and magnetic data were made. First of them was a structural analysis that allowed us to suppose that N–S direction basement graben structure locates close to the western Sambia in the Gdansk Bay. The second approach was a profile data wavelet transformation. This profile had a latitudinal direction and passed trough possible source area. In the result, at least three N‐S faults of different morphology from strike‐ slip up to normal connected with the basement and probably with deeper structures were revealed. We think that all Kaliningrad earthquakes occurred on one of these faults.

111444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

TTTHHHEEE NNNEEEWWW SSSWWWEEEDDDIIISSSHHH NNNAAATTTIIIOOONNNAAALLL SSSEEEIIISSSMMMOOOLLLOOOGGIIICCCAAALLL NNNEEETTTWWWOOORRRKKK

Bödvarsson R. Uppsala University; [email protected]

Over the last few years, 60 new, permanent, digital, broadband seismological stations have been deployed in Sweden, from Lannavara in Lappland in the North to Skåne in the South. The network operates largely automatically, and is now essentially complete in terms of number of stations. The primary objective of the network is to monitor local seismicity. With the current station spacing of about 100 km, completeness down to magnitude 0 is assured within the network. This magnitude corresponds to very small movements, for example to a motion of 0.01 mm over a fault area with a radius of about 50 m. Several hundred Swedish earthquakes are detected every year. Only a few (5 to 10) of these are so large that they are felt by people living close to the epicentre. While Sweden is a low seismicity area, the high sensitivity of the system means that ongoing deformation processes in the crust can be monitored in detail. As a larger data set is gradually acquired, it will also be possible to use information from these events to elucidate structures within the Swedish crust. In addition, the network records signals from larger distant (teleseismic) earthquakes, and also regional events of sufficient magnitude. These data are analysed to reveal details of the structure within the crust and upper mantle below the recording stations.

111555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

NNNEEEWWW PPPEEERRRSSSPPPEEECCCTTTIIIVVVEEESSS OOOFFF MMMOOONNNIIITTTOOORRRIIINNNGGG OOOFFF TTTHHHEEE SSSEEEIIISSSMMMIIICCCIIITTTYYY OOOFFF GGGUUULLLFFF OOOFFF GGGDDDAAANNNSSSKKK AAANNNDDD AAADDDJJJAAACCCEEENNNTTT AAARRREEEAAASSS

Dębski W., Wiejacz P., Suchcicki J., Wiszniowski J. Institute of Geophysics, Polish Academy of Sciences, ul. Ksiecia Janusza 64, 01‐452 Warsaw, Poland; [email protected]

Abstract

The occurrence of two relatively large (Mw = 5.0, 5.3) earthquakes that occurred on September 21, 2004 at the Sambia Peninsula (Kaliningrad region, Russia) has renewed an interests in the seismicity and seismic hazard of the region which always was considered to be very small and weak. Actually, some previous studies reveled a small seismic activity in the Gulf of Gdansk area. For example, such events were (and are) sporadically recorded by the closest station SUW or the CZAJ stations that operated between July 1997 and August 1999. However, the recorded events were very small (M< 3) and the closest seismic stations (or very powerful array like NORESS) where too far away to assure the completeness of recording seismicity. For this reason, to have a better insight into seismicity of the southern Baltic area an afford has been undertaken by IGF PAS to establish a new seismic station in the area – on the tip of the Hel Peninsula at the geophysical observatory of Polish Academy of Sciences. This attempt and the first results are shortly discussed in the context of the previous observation of seismicity of northern Poland.

111666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSOOOUUURRRCCCEEE PPPAAARRRAAAMMMEEETTTEEERRRSSS OOOFFF KKKAAALLLIIINNNIIINNNGGGRRRAAADDD EEAAARRRTTTHHHQQQUUUAAAKKKEEESSS

Domański B. and Dębski W. Institute of Geophysics, Polish Academy of Sciences, ul. Ksiecia Janusza 64, 01‐452 Warsaw, Poland; [email protected]

Abstract

Two surprising earthquakes hited the Sambia Peninsula (Kaliningrad region) on September 21, 2004. They were astonishing not only because of their occurrence, large magnitudes, almost no aftershock sequence (only one recorded instrumentally by the closest SUW station), but also very particular characteristics revealed by the spectral analysis of the recorded seismograms. Both events were right‐lateral, deep earthquakes with very well visible complicated radiation pattern and surprisingly large high frequency content of seismic traces recorded by station closed to the rupture plane with respect to the classical source models like Brune or Madariaga’s models. These two features explicit influence estimators of such parameters as, for example, magnitude or the source radius calculated for a single station and cause some scattering of obtained values. The interpretation of the high frequency content of some spectra is still not clear. One of the possible mechanism leading to so unexpected large high frequency part of the spectrum is hypothesis, that due to a complex geology in the source area a guideline is formed of dimensions appropriate for an efficient transporting the high frequency energy at large distances. An another possible explanation is a hypothesis on a complexity in the source generation process. Currently, with the available seismic data none of the hypothesis like that can not be easily ruled out.

111777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

RRREEECCCOOONNNSSSTTTRRRUUUCCCTTTIIIOOONNN OOOFFF SSSTTTRRREEESSSSSS FFFIIIEEELLLDDDSSS IIINN AAADDDJJJAAACCCEEENNNTTT RRREEEGGGIIIOOONNNSSS SSSEEEPPPAAARRRAAATTTEEEDDD BBBYYY SSSEEEIIISSSMMMIIICCCAAALLLLLLYYY‐‐‐AAACCCTTTIIIVVVEEE FFFAAAUUULLLTTTSSS

Galybin A. N.1 and Mukhamediev Sh. A.2 1Wessex Institute of Technology, Ashurst, Southampton, SO40 7AA, UK; [email protected] 2Institute of Physics of the Earth, Bolʹshaya Gruzinskaya ul. 10, Moscow, 123995, Russia; [email protected]

PRELIMINARIES. Identification of stress fields in the earth’s crust is an important problem of geophysics and rock mechanics. The methods for in‐situ stress measurements have recently been reviewed in the Special Issue of the IJRMMS (see for instance, [1–3] and other papers in this issue). However most of these methods provide small scale measurements and therefore a sufficient number of single measurements is required for reliable transition to the regional stress, which necessitates the development of adequate mathematical modelling. Conventional modelling is based on classical formulations of boundary value problems of elasticity. In these approaches, stress fields are calculated for diverse boundary conditions defined on the margins of the region in order to fit the experimentally observed orientations of principal stresses inside the region, see, for instance, [4–6]. The drawbacks of such approaches have been discussed by Mukhamediev et al [7] who also suggested three alternative methods for determining the elastic state of stress in relatively stable blocks of the lithosphere. All of the methods use, as input information, the experimental data on the stress orientations. The latter is mostly obtained from the analysis of seismic data. Thus, the current 2005 release of the World Stress Map database [8] contains 15,969 data points (mostly on stress orientations), 10,619 of which have been collected from earthquake focal mechanisms. The other stress indicators include well bore breakouts and drilling induced fractures (3365), in‐situ stress measurements (overcoring (611), hydraulic fracturing (349), borehole slotter (33)) and other methods. All data are quality ranked according [9] and the following quality categories have been assigned: A (6%), B (7%), C (62%), D (16% ), E (7%). The highest quality is A; the A‐ quality data are believed to record the orientation of stresses to within ±10°–15°, the B‐quality data to within ±15°–20°, and the C‐quality data to within ±25°. D‐quality data are questionable, E‐quality data are not reliable. It should be noted that, in general, seismic data is of better quality than data obtained by the other methods. This paper presents a method for identification of elastic stresses in adjacent regions by employing the data on stress orientations (principal directions of the stress tensor). Principal directions on both sides of the interface between the domains represent two boundary conditions, the other two are provided by assuming continuity of the stress vectors across the interface. The formulation does not require the knowledge of stress magnitudes, and therefore the complete stress tensor can only be determined with some degree of arbitrariness. For considered examples, the analysis shows that two real parameters remain undetermined in the complete solution for stress components. However maximum shear stress (acting in‐plane) contains one unknown multiplicative constant and the field of stress trajectories is unique. Many observations show that one of the principal directions is usually sub‐vertical [10]. This allows one to separate out‐of‐plane and in‐plane stress fields. In many cases the earth’s curvature can be neglected in the first approximation, which produces an error comparable with the errors due to data scattering. Therefore we further consider the in‐plane stress tensor that can

111888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

be characterised by three independent components or by two principal stresses σ1≥σ2, and the principal direction, ϕ, which is the direction σ1 with respect to a reference coordinate system The main feature of the data is that stress orientations are mostly available in a relatively narrow zones associated with the region boundaries or major faults. This necessitates combining two approaches recently suggested in [11, 12] for solving elastic problems in which stress orientations are known, while stress magnitudes remain unknown everywhere. First approach deals with the boundary value problem, BVP, formulated in terms of stress orientations. This BVP has no unique solution in general case, however, the number of independent solutions can be identified from the distribution of stress orientations along the entire boundary. Then the total solution of the BVP is constructed as a linear combination of independent solutions and hence it contains several arbitrary real constants. The arbitrary constants cannot be identified from stress orientations alone but they can be determined if stress magnitudes are known at some locations. Second approach is applied when stress orientations are known at discrete points located within the considered domain. In this case the problem does not belong to any type of BVPs and its solution is sough as a linear combination of basis functions with unknown coefficients. These coefficients are determined by matching the observed and predicted data. In this paper this approach is applied to the case when two blocks of the lithosphere are separated by the major fault. An example for the Sunda trench, which is the boundary between Indo‐Australian and Eurasia plates, is presented further. MATHEMATICAL MODEL. Formulation of plane boundary problems of elasticity in terms of stresses includes two equations of equilibrium, condition of compatibility and two boundary conditions posed on the entire contour of each considered region. For plane isotropic elastic domains, the Kolosov‐Muskhelishvili solution given in terms of complex potentials is valid (no body forces) σ + σ σ − σ 11 22 = P(z, z )= Φ ()z + Φ ()z , 22 11 + iσ = D()z, z = zΦ′()z + Ψ ()z (1) 2 2 12 Here Φ(z) and Ψ(z) are holomorphic functions of complex variable z=x1+ix2; P()z, z , D()z, z are stress functions that represent mean stresses and stress deviator

respectively; the over‐bar stands for complex conjugate, thus z = x1 − ix2 . Stress vector on a contour Γ is expressed as follows dζ P()ζ + D()ζ = N()ζ + iT ()ζ , ζ ∈ Γ (2) dζ Hereafter a single variable is used as the argument of a function in order to emphasize its boundary value. N and T are normal and shear components of the stress tensor on G. The following boundary conditions are accepted ⎛ 2iϕ± ()ζ ± ⎞ + + − − Im⎜e D ()ζ ⎟ = 0, N ()ζ + iT ()ζ = N ()ζ + iT ()ζ , ζ ∈ Γ (3) ⎝ ⎠ Hereafter “±” denote the boundary values obtained by approaching Γ from different domains Ω± (domain Ω± lies on the left (right) of the contour respectively). The mathematical problem consists in the determination of complex potentials and stress functions by boundary conditions (3). As soon as potentials are found, the stress fields (i.e., stress functions and stress components) in both exterior and interior domains can be determined by formulas (1). It is proved [13] that this problem has solutions if and only if 2Κ≥0, where –2πΚ is the increment of ϕ+ (ζ) − ϕ− (ζ) after the complete traverse of Γ in positive (counter‐clockwise)

111999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

direction. The number of linearly independent solutions is 2Κ+1 and the general solution may include up to 4K+2 real arbitrary constants that are free parameters. It is evident that the structure of data on stress orientations determines the number of free parameters that cannot be identified from the analysis. These can only be found if additional information involving stress magnitudes is known at some locations. Following the methodology proposed earlier (see Galybin & Mukhamediev, 2004) we further reduce (3) to the following minimisation problem 2iϕ Im⎜⎛e j D()z , z ⎟⎞ → min, with constrain (2) (4) ⎝ j j ⎠

Here ϕj are principal directions at data points zj, symbol ||…|| stands for the Euclidian norm. Solution for stress functions is sought as a linear combination of bi‐holomorphic functions qk(z) and harmonic functions pk(z) with unknown complex coefficients ck (k=1..n) as follows 8n+3 ⎛ F ()z, z ⎞ ⎛q± ()z, z ⎞ D± ()z, z = aF ()z, z , P± ()z, z = b + aF ()z, z , ⎜ 1 ⎟ = c ⎜ k ⎟ (5) 1 2 ⎜ F ()z, z ⎟ ∑ k ⎜ ± ⎟ ⎝ 2 ⎠ k=0 ⎝ pk ()z, z ⎠ Substitution of (5) into (4) followed by the application of the collocation method results in a system of linear algebraic equations for the determination of 8n+4 unknowns c0…c8n+3. The system is, in general, overspecified, it has 8n+4 unknowns and the number of equations, N, is equal to N=N+ +N− +NΓ+1, where N+ =number of data in Ω+; N−=number of data in Ω− Γ ,N =number of collocation points, ζm, on the contour. One more equation is added, it expresses the fact that the stress deviator cannot be found uniquely and it should be normalised in order to find non‐trivial solution of the system. Normalization is chosen such that the average value of D over all data is unity Therefore, the right‐hand sides for all equations (except for the last one) is zero bj=0, j=1…N‐1. Solution of the system is found by the least squares method. EXAMPLE. The Sunda trench that represents the boundary between Indo‐Australian, plate IAP, and Eurasian plate, EAP has been considered as the only boundary in the problem. Other boundaries have been excluded from the consideration because the data (shown in the left figure) demonstrate that the difference in stress orientations is not pronounced on the other boundaries. It has been found that among 167 data of A‐C quality 47 points lie within IAP and 120 in EAP. The number of collocation has been chosen the same NΓ=N=167. The field of stress trajectories in the figure below acknowledges the sharp change of stress orientations on different sides of the Sunda trench. Trajectories approach the trench almost perpendicular from EAP and tend to be parallel to it in IAP. This indicates low shear resistance along the trench. Maximum shear stresses are presented in the right figure together with epicentres of resent earthquakes. Analysis has shown that principal stresses σ1 and σ2 are closed to each other in the area adjacent to the Sunda trench (light zone in the figure). This is in accordance with the stress regime in this area. CLOSURE. This article presents a method for reconstruction of stress fields in adjacent regions separated by faults. The method suggests utilisation of seismic data for determination of stress orientations that are further used as input. It is remarkable that no information on boundary stress magnitudes is required, which allows for unique determination of the stress trajectories pattern, while the complete recovered stress tensor may contain up to two free parameters. The example for the Sumatra region is in agreement with the characteristics of the ‐served stress field.

222000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure. Pattern of stress trajectories (left) and normalised maximum in‐pane shear stresses (right) near Sumatra (background map supplied by the WSM project). Asterisks indicate locations of resent earthquakes.

ACKNOWLEDGMENT. The work is partly supported by the Russian Foundation for Basic Research (grant No 06‐05‐65245)

REFERENCES: Fairhurst C. 2003. Stress estimation in rock: a brief history and review. IJRMMS. 40: 957‐973. Hudson JA, Cornet FH, & Christiansson R. 2003. ISRM Suggested Methods for rock stress estimation— Part 1: Strategy for rock stress estimation. IJRMMS. 40: 991–998 Bérard T & Cornet FH. 2003. Evidence of thermally induced borehole elongation: a case study at Soultz, France. IJRMMS 40: 1121–1140. Cloetingh S. & Wortel R., 1986. Stress in the Indo‐Australian plate. Tectonophysics, 132: 46‐67. Coblentz D.D., Sandiford M., Richardson R.M., Zhou S. & Hillis R., 1995. The origins of the intraplate stress field in continental . Earth Planet. Sci. Lett.,, 133, 299‐309. Zhang Y., Scheiber E., Ord A. & Hobbs B.E., 1996. Numerical modelling of crustal stresses in the eastern Australian passive margin. Aust. J. Earth Sci., 43: 161‐175. Mukhamediev, Sh.A., Galybin A.N. and Brady, B.H.G. 2006. Determination of stress fields in elastic lithosphere by methods based on stress orientations. IJRMMS 43 (1): 66‐88. Reinecker, J., Heidbach, O., Tingay, M., Sperner, B., & Müller, B. (2005): The release 2005 of the World Stress Map (available online at www.world‐stress‐map.org). Zoback, M. L. and Zoback, M. D. 1989. Tectonic stress field of the conterminous United States, Mem. Geol. Soc. Am. 172: 523‐539. Zoback, M. L. et al. 1989. Global patterns of tectonic stress. Nature 341: 291‐298. Galybin, A.N. & Mukhamediev, Sh.A. 1999. Plane elastic boundary value problem posed on orientation of principal stresses, J. Mech. Phys. Solids. 47, 2381‐2409. Galybin, A.N. & Mukhamediev, Sh.A. 2004. Determination of elastic stresses from discrete data on stress orientations. Int. Journal of Solids and Structures. 41 (18‐19), 5125‐5142. Galybin, A.N. & Mukhamediev, Sh.A. 2006. Integral equations for elastic problems posed in principal directions: application for adjacent domains. In C. Brebbia (ed.), Boundary Elements 28, WIT Press, Southampton, UK: 51‐60.

222111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

CCCOOORRRRRREEELLLAAATTTIIIOOONNNSSS OOOFFF MMMAAAGGGNNNIIITTTUUUDDDEEE AAANNNDDD FFFEEELLLTT‐‐‐AAARRREEEAAA FFFOOORRR EEEAAARRRTTTHHHQQQUUUAAAKKKEEESSS IIINNN TTTHHHEEE FFFEEENNNNNNOOOSSSCCCAAANNNDDDIIIAAANNN SSSHHHIIIEEELLLDDD///EEEAAASSSTTT EEEUUURRROOOPPPEEEAAANNN PPPLLLAAATTTFFFOOORRRMMM

Gregersen S.1, Husebye E.2, Mantyniemi P.3 1GEUS, Ostervoldgade 10, DK‐1350 Copenhagen K, Denmark, [email protected], 2University of Bergen, Norway 3University of Helsinki, Finland

Magnitudes of historic, pre‐instrumental earthquakes are estimated from observed intensities and felt areas. For a start, a basic requirement is access to macroseismic and instrumental observations for establishing meaningful correlations between modern digital magnitudes and felt areas for various intensity levels. This has to be established region by region. Commonly a log‐linear relation is established between instrumental magnitudes and size of felt area for intensity level 3.

intensity 3

7

6 5

4 int3KaDS 3 int3Nor

(area) log 2 1

0 0123456

magnitude mL

Here int3KaDS means: intensity 3 in Kaliningrad, Denmark and Sweden and int3Nor means: intensity 3 in Norway.

We test also for intensity 4 felt area as the macroseismic parameter.

222222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

intensity 4

6 5 4 int4KaDS 3 int4Nor

log (area) 2 1 0 0123456 magnitude mL

Here int4KaDS means: intensity 4 in Kaliningrad, Denmark and Sweden and int4Nor means: intensity 4 in Norway.

Recently we have established an interactive graphic procedure for estimating sizes of felt areas. This procedure has been tested on recent earthquake observations in with satisfactory results. The set of data points has just been extended, so that judgement becomes better for larger magnitudes. New important key points in the correlation between modern digitally‐determined magnitudes and felt areas have namely become available from the Kaliningrad earthquakes in 2004. They can be used together with smaller events to scale old earthquakes. This gives improved judgement of the larger‐than‐usual earthquakes like the Oslo earthquake 1904, which is very similar to the largest Kaliningrad earthquake 2004 of Lg‐ wave magnitude 5.2.

222333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

NNNOOORRRWWWEEEGGGIIIAAANNN NNNAAATTTIIIOOONNNAAALLL SSSEEEIIISSSMMMIIICCC NNNEEETTTWWWOOORRRKK RRREEECCCOOORRRDDDIIINNNGGG EEEVVVEEENNNTTTSSS IIINNN TTTHHHEEE BBBAAALLLTTTIIICCC CCCOOOUUUNNNTTTRRRIIIEEESSS

Havskov J. Department of Earth Science, University of Bergen, Allegaten 41, 5007 Bergen, Norway, [email protected]

In Norway, the University of Bergen (UiB) operates the Norwegian National Seismic Network (NNSN) consisting of 29 seismic stations where 8 have broad band sensors and the rest short period sensors. Nearly all stations operate with the public domain software SEISLOG and SEISAN developed in Bergen. NORSAR operates 3 seismic arrays and one seismic station (Figure 1). Data from all stations in Norway and some stations in surrounding countries are combined to make a Scandinavian earthquake catalog with special emphasis on Norwegian areas.

Figure 1. Norwegian seismic stations. UiB operates the 29 stations in the National Seismic Network (NNSN) and NORSAR operates the 3 arrays and the station JMIC.

The catalog contains more than 40000 events for the period 1800 to 2007 in the area shown in Figure 1 (NNSN prime area). For 2006, 1934 events were located in the prime area, which also includes a few events (explosions) in the Baltic countries (Figure 2).

222444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

2500 of the 40000 events are in the Baltic area (Figure 3).The majority of Baltic events in the NNSN data base have been recorded only with the NORSAR and Finish arrays due to the large distance to the events and the emergent nature of the signals. The first events in our data base are from 1963 while the majority of the events are after 1990, when the integrated NNSN started operation and integration with NORSAR data was improved. It seems that NNSN without the Scandinavian arrays is capable to record events down to magnitude 2.5 in the Baltic countries although many larger events only are seen on the arrays, probably due to the emergent nature of the signals.

Figure 2. Epicentre distribution of events analyzed and located in 2006. Earthquakes are plotted with grey symbols and probable and known explosions with whitecircles. 37% of the events were identified as probable or confirmed explosions.

222555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 3. Seismic events from the Baltic area as recorded by Norwegian stations. The time period is from 1964 to 2006.

From Figure 3, the clustering of events in the mining areas is clearly seen. Probably most of the events are explosions and NNSN is trying to identify them by location and time of day distribution. Of the 2500 events, 1934 were identified as probable explosion. The remaining probable earthquakes are seen in Figure 4. From the clustering it is clear that most of them are explosions.

Figure 4. Seismic events from the Baltic area as recorded by Norwegian stations. The time period is from 1964 to 2006.

222666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

It is clearly a challenge to identify real earthquakes in the Baltic countries, at least as seen from the Scandinavian seismic network, so a good knowledge and reporting of Baltic explosion is essential to get a real overview of Baltic earthquake activity. All the data stored in the NNSN database is also available to the public via Internet, e‐ mail or manual request. The main web‐portal for earthquake information for NNSN is www.skjelv.no. The software used in NNSN is freely available with source code at http://www.geo.uib.no/seismo/

222777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

LLLIIITTTHHHUUUAAANNNIIIAAANNN SSSEEEIIISSSMMMIIICCC NNNEEETTTWWWOOORRRKKK ––– CCCUUURRRRRREENNNTTT SSSTTTAAATTTUUUSSS AAANNNDDD PPPEEERRRSSSPPPEEECCCTTTIIIVVVEEESSS

Lazauskienė J. Lithuanian Geological Survey, Konarskio 35, Vilnius, Lithuania, [email protected]

The territory of Lithuania and adjacent areas in the Baltic Region, located in the south‐western margin of the East European Craton, could be considered as a region of a low seismicity. Nevertheless, historical records reveals several rather prominent seismic events through the ages. The first instrumental seismic observations started in 1970 in Vilnius seismic station and in 1999 four seismic stations, comprising the Seismic Monitoring System (SMS) of Ignalina Nuclear Power Plant (INPP) were installed. The same year with establishment of SMS, Geological Survey of Lithuania (LGT) launched the project of the “Seismological monitoring of Lithuania”, aimed to process, analyze and store seismological data of SMS. The data of seismic observations were systematically collected, archived, analyzed and interpreted by specialists of the Department of Bedrock Geology of LGT in collaboration with the specialists of INPP and different international organizations. Processed seismic data are provided to the international seismological centres and institutions of neighbouring countries, the annual seismological bulletins are compiled and presented to the public every year. Currently only these four seismic stations, owned by INPP and located in the Northeastern part of Lithuania are operating in the country (Fig. 1). Still, the seismic network of INPP is not capable to cover the whole territory of Lithuania with trustable resolution (e.g. the fact, that no quarry blast (distance to INPP ~200 km) was recorded by seismic monitoring system (SMS) of INPP evidences this issue) and the seismo‐tectonic structure of Lithuania and adjacent territories is not completely understood yet – the earthquakes of moderate size in the Kaliningrad district of Russian Federation 21 September 2004 caused a seismic shaking throughout all territory of Lithuania, provoking the great concern of mass media and public and proving the necessity of the state strategy of seismic investigations and establishment of the national network of seismic stations, covering the entire territory Lithuania. Following the Kaliningrad events, September 2006 the Government of the Republic of Lithuania approved „The program of the assessment of the seismicity of the territory of Lithuania for years 2007– 2010“.The program is aimed to ensure the assessment of the structure of the territory of Lithuania in terms of the seismic activity, the persistent seismologic monitoring of, establishment and maintenance of the national seismic network in Lithuania and compilation of the map of seismic hazards of the country as a final result. Joining and the data exchange the other international seismic monitoring networks and participation in the international projects is one of the tasks of this program. LGT, as the major body, responsible for the implementation and the supervision of the aforementioned program, started the second project of Seismological Monitoring of Lithuania at the end of year 2006. The project continues the activities, started by the first stage of the project during the 1999–2005 resulting in collecting, processing and analyzing the data of Seismic Monitoring system of INPP, latter on, accomplished by the data of the planned newly established broadband seismic stations. Seismic data registered by the other systems of seismological monitoring of Baltic region will be analyzed too. The other of the means of the implementation of the aforementioned seismological program is the project of the “Passive seismic monitoring of the territory of Lithuania“, that

222888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

has been started the very end of year 2005 in the framework of the international Passive Seismic Experiment PASSEQ 2006–2007 (PASSEQ 2006–2007 – the international research project with more than 10 countries participating and ~150 seismic stations located in German, Czech Republic, Poland and Lithuania). The major aim of the project is to investigate the deep structure of the lithosphere of the Central and Precambrian East European Craton utilizing records of teleseismic events, to register the regional earthquakes in order to understand the seismo‐tectonic processes of the territory. In a course of year 2006 the deployment and installation of 26 temporary seismic stations has been fulfilled in the territory of Lithuania (Fig. 2), the regular maintenance and observation of the stations is carried out. Important part the project contains the downloading, copying and archivation the raw digital data and providing them for the international project co‐ordinators. It is foreseen to keep the recording and observation time till the end of 2007. The new data, obtained in a course of the project are requested in order to define the structure of the major tectonic zones of Lithuania, for the prognosis of the geodynamic activity of the crust. According to the means of the implementation of the programme of the seismicity assessment of Lithuania, installation of two new broadband seismic stations, associated to the GEOFON network, is foreseen for year 2008–2009. The first step towards the establishment of the Lithuanian national seismic network was preparation of the investment projects for the two new seismic stations installation that were submitted to the Lithuanian Government and the Ministry of Environment in 2007. As the second step, the Memorandum of Understanding for Seismological Coordination, Cooperation, and Collaboration among the LGT and the Geoforschungszentrum, Potsdam (GFZ) has been signed June 2007. In case of the provided state financing two new broadband sesmic stations, associated to the GEOFON network, would be constructed and installed in the Western and Central parts of Lithuania (Fig. 1) in 2008–2010. The vault type seismic stations would be situated ~50 and ~200 km away from the coast of the Baltic Sea on soft sediments (Quaternary glacial loam and sandy clay deposits), creating the instrumental bunker. Two STS‐2 sensors would be mounted at a 5–6 m depth from the ground surface. LGT and the GFZ will operate jointly those two seismic stations, cooperate in data processing and provide the equal and open access to all data of the GEOFON network or all broadband stations operated by the LGT. The basic equipment (seismometer and digitizer) of the seismic stations, construction of the seismic vault and the provision of the data communication and transfer would be provided by LGT while required equipment orders, an additional equipment (e.g. Seiscomp station processor, seismometer shielding, etc.), the station installation and training for operation and maintenance would be ensured by GFZ partners. Currently the site(s) selection of two broad bend seismic stations in the Western and Central Lithuania is carried out be specialists of LGT. The negative screening of the territories of interest was performed considering the land ownership and use, infrastructural, geographical, geological (type pf sediments), hydrogeological (level of groundwater) etc. factors to reveal the territories where installation of the seismic stations is unfavourable. 36 potentially suitable candidate sites, selected on a basis of ten major criteria (such as: the land ownership; access to station; distance form the railway and major roads, the level of ground water, electricity and internet connection, distance to a lakes and local roads, type of underlaying sediments, relief etc.) predominantly located in Western Lithuania, has been investigated. Potentially suitable sites were evaluated in order to select few most perspective ones. In respect to the requirements of the seismic stations site selection and the negative screening, few candidate sites, located in Paragiai Manor (Akmenė region), showing the best fit

222999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

to the requirements e.g. described by favourable infrastructural characteristics and good topographical, geological, hydrogeological conditions, could be considered as the most perspective candidates in the Western Lithuania, while the site, located in the close vicinity of the Administration of Krekenava Regional Park, could be regarded as the candidate for the Central Lithuania. Still, more detail investigations of the potential sites are required. The evaluation of seismic noise sources is foreseen to be carried out in the nearest future.

Fig. 1. Planned national seismic network in Lithuania. Bounded triangles – four existing seismic stations of Ignalina NPP, red squares – two new GEOFON network broad bend seismic stations foreseen to be installed in 2008–2010.

Fig. 2. Location of 26 temporary seismic stations of the PASSEQ‐ 2007 project installed in the territory of Lithuania.

333000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEELLLEEENNNAAA ––– AAA SSSOOOFFFTTTWWWAAARRREEE FFFOOORRR NNNEEEAAARRR RRREEEAAALLL‐‐‐TTTIIIMMMEEE DDDAAAMMMAAAGGGEEE EEESSSTTTIIIMMMAAATTTIIIOOONNN AAANNNDDD DDDAAAMMMAAAGGGEEE SSSCCCEEENNNAAARRRIIIOOOSSS

Lindholm, C. D.1, 3, Molina S.2, 3, Bungum H.1, 3 and Oye V.1, 3 1NORSAR, Kjeller, Norway 2University of Alicante, Alicante, Spain 3International Centre for Geohazards, Oslo, Norway

Abstract

Loss estimation studies have traditionally been using individual scenario earthquakes as the basis for planning and decision making. Existing knowledge of regional geology and seismology have, in the same context, been used to generate maps with estimated intensities or accelerations which, in combination with other factors, have been used for estimating damage to structures and lifelines, and impacts on population. Nowadays, the advent of high speed computing, satellite telemetry and Geographic Information Systems (GIS) have made it possible to generate rapid loss estimates for multiple earthquake scenarios, and to provide a nearly unlimited mapping capability. Perhaps most importantly, it is now within reach to develop damage and loss estimates for an actual earthquake in near real‐time based on the source parameters of an event, with magnitude and location as a minimum. We have investigated the capabilities of a recently developed seismic risk tool (SELENA) for being used as a near real‐time damage estimation software. From the obtained results we have concluded that the variability in input parameters can be introduced in a logic tree with 27 branches, covering scenarios that are computed fast enough to be considered ‘near real time’. Although more testing needs to be conducted before any near‐real‐time implementation of SELENA, the tool is considered very promising. Introduction

The developments within earthquake engineering over the past decade (e.g. Fajfar, 1999) have significantly improved the capacity to analytically and empirically evaluate building performance during earthquake shaking. This has facilitated earthquake damage scenarios (Ordaz et al., 2000; Bommer et al., 2002; Carvalho et al., 2002, Erdik et al., 2004; Molina and Lindholm, 2005; among others) for single structures as well as for larger number of distributed buildings (such as cities and communities). Mainly because the damage scenario results can be presented in intuitive figures and maps these developments have provided the scientists with tools that can be useful 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 earthquake risk evaluations, and for this reason many new 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 (automatic and semi‐automatic) determination of hypocenters and even fault ruptures, following a large earthquake, near‐real‐time shake maps can be computed and offered to emergency and civil protection agencies as well as to the the general public through web portals (e.g. Wald et al., 1999). A further step in the same direction is to compute (at different levels of resolution)

333111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

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 important is that after a large and damaging earthquake a damage information overview is often 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 (Wyss, 2006).

Methods and Approaches 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 and ground shaking in terms of intensity (MMI) were defined and damage probability matrices (DPM’s) were developed. • An analytical approach was used in the HAZUS project (FEMA, 1999, 2004), consuming considerably more resources and focusing on an engineering‐wise characterization of a wide range of buildings. For 147 different building types (in terms of structural design, building regulatory regime, maintenance, age etc.) building capacity curves and associated shaking vulnerability functions were defined so that damage could be estimated quantitative from ground motion estimates. These two approaches, using Intensity based Damage Probability Matrices and vulnerability functions based on ground acceleration, respectively, are still followed in various approaches today. In a city located in a high seismicity region predefined scenarios may be a useful approach when the causative faults are (believed to be) known in detail, in particular since this allows for testing of several magnitude scenarios so that the exposure variation can be explored in more detail. In regions with low‐to‐intermediate seismic activity or in regions with very complex tectonics a reliable mapping of future large fault ruptures may be impossible, and in such cases near‐real‐time ground shaking maps can be used for estimating damage. Such maps can be estimated from location and magnitude estimates alone, possibly combined with structural‐geological information, but the maps will be much more reliable if they are also based on real‐time strong‐motion data (Wald et al., 1999). The method followed in the approach presented here is based on the HAZUS methodology (FEMA 1999, 2004). With SELENA (SEimic Loss EstimatioN using a logic tree Approach; Molina and Lindholm, 2006) the HAZUS methodology has been extended to a logic tree computation and developed in a MATLAB environment providing the user with full flexibility with respect to input and presentation of results. A rapid earthquake damage processing scheme is outlined in Fig. 1. The seismological observatory will be the first entity to detect the earthquake, and an automatic location process is initiated. After the first determination of epicenter and magnitude the results may be provided for the first near‐real‐time ground motion scenario while at the same time the computation process to define the source more precisely continues. Once the first ground motion maps are generated, these are handed over to the SELENA process where two approaches may be followed in parallel as shown in Fig. 2. Depending on the type of

333222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

information provided for SELENA, one may run one of the indicated processes or both in parallel.

Static information Dynamic information Fig. 1. Simplified processing scheme Early epicenter and for a rapid damage information Building inventory magnitude and damage models determination system. The components in the Population inventory Lifelines inventory red dotted box indicates the core and damage models of the damage computations. ……… Determine earthquake source parameters These engineering computations

depend on input information that Compute damage and Calculate regional is both static and dynamic. See casualties ground motion Fig. 2 for more details.

Compare observed damage Compare ground with ???predicted motion with predicted

Output: Maps and Graphs

The SELENA code is based on a parameterized building inventory which has been pre‐ compiled. The approach is often called the ‘capacity‐spectrum method’, because it combines the ground motion input in terms of response spectra (spectral acceleration versus spectral displacement) with the building specific capacity curve. The philosophy is that any building is structurally damaged by its permanent displacement and not by the acceleration by itself. The building capacity curve is defined through three control points, Design, Yield and Ultimate capacity (Fig. 3). Up to the yield point, the building capacity curve is assumed to behave elastically linear. From the yield point to the ultimate point, the capacity curve changes from an elastic to a fully plastic state (curved form), and the curve is assumed to remain fully plastic past the ultimate point (linear form). The structural damage states are (as in most other proposed schemes) divided into four damage states, slight, moderate, extensive and complete. A detailed description of these damage states are available from many sources, and Fig. 4 shows an example of the damage state probability curves that are used as basis for the final damage estimation maps.

Defined fault model for No clear rupture well established source model: Shake maps

Fig. 2. Delineation of the damage Search library for damage computation along two scenario based on source Compute damage to branches depending on input model close to the the physical established (magnitude, environment based on form. A) Using a library of shakemaps location etc. precomputed scenarios if the source is well defined. B) Using

Merge and interpolate near‐real‐time computation of damage statistics shake maps in other situations.

Compute losses and casualties

Output: Maps and Graphs

333333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

+1 σ

SPECTRAL ULTIMATE

Spectral Spectral Acceleration LOAD CAPACITY Median Au :

Ay : YIELD CAPACITY -1 σ Ad : DESIGN CAPACITY

Dd Dy Du Spectral Displacement Fig. 3. The principle of the building specific capacity curve intersected by the ground motion load spectrum.

Results To test the procedures above we used data from the city of Oslo, and the test was conducted for a deterministic scenario where we varied the source magnitude (and thereby the fault rupture length), the soil amplification, the attenuation relations, the vulnerability functions of the structures and the economic loss models. The city of Oslo was chosen as a test case because all the inventory information was already in place; however, since the present study is investigating the capacities of a particular software package (SELENA) the test site as such is unimportant.

1 t e h t g a li r e e S v d i o s n M e t te x le E p m o C age probability m Da

0 Displacement Fig. 4. The derived vulnerability (or fragility) probability functions.

We have investigated the SELENA performance for a deterministic and a probabilistic scenario. However, since only deterministic scenarios are relevant for real‐time computations only that aspect has been documented in the following. The results obtained from these tests are outlined in Figs. 5, 6 and 7 where we have investigated the difference in the damage predictions using the different branches in the logic tree computations. The computation for a city is subdivided so that it provides damage for the smaller city areas, and the present study has (for practical reasons only) used census tracts as the smallest entity.

333444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Differences in losses

200.0 180.0 Diff 1-3 160.0 Diff 1 -5 140.0 Diff 3_5 120.0 100.0 80.0

% Difference 60.0 40.0 20.0 0.0 1 5 9 131721252933374145495357616569737781 Census Tract Number

Fig. 5. The observed differences in losses in each census tract when using a) the difference in % between the results obtained only with one magnitude 6.0 earthquake and 3 earthquakes (5.5, 6.0, 6.5) weighted (0.2, 0.6, 0.2); b) the difference between one magnitude 6.0 earthquake and 5 earthquakes (5.5, 5.75, 6.0, 6.25, 6.5) weighted (0.1, 0.1, 0.6, 0.1, 0.1); and c) the difference between 3 earthquakes (5.5, 6.0, 6.5) weighted (0.2, 0.6, 0.2) and 5 earthquakes (5.5, 5.75, 6.0, 6.25, 6.5) weighted (0.1, 0.1, 0.6, 0.1, 0.1).

Firstly, Fig. 5 investigates the number of necessary magnitude branches by comparing the results for 1 and 3, for 1 and 6 and for 3 and 6 branches, respectively. It is obvious that the differences are large and that they vary between the census tracts since some census tracts are closer to the source and are hence much more affected by changes in the rupture length (as function of magnitude). However, when disregarding the census tracts that are most unstable, the general trend is that three different branches on magnitude yield results that are fairly similar to the results when using six branches. These results in Fig. 5 are important in the sense that they justify limitations in the number of branches.

180.00 dif 1 y 9 160.00 dif 1 y 27 dif 1 y 81 140.00 dif 1 y 162 120.00 dif 3 y 9 dif 9 y 27 100.00 dif 27 y 81 dif 81 y 162 80.00 % Difference 60.00

40.00

20.00

0.00 0 1020304050607080 Census Tract

Fig. 6. Difference between predicted losses in each census tract for different number of branches in the logic tree computation. Legend example explanations: 1–9: 1 magnitude, 1 soil, 1 attenuation, 1 capacity‐fragility, 1 economic model versus 3 magnitude, 3 soil, 1 attenuation, 1 capacity‐fragility, 1 economic model model branches; 1–162: One versus 3 magnitude, 3 soil, 3 attenuation, 3 capacity‐fragility, 2 economic model branches.

333555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Fig. 6 addresses a situation where branching is done simultaneously over magnitude, soil amplification, strong‐motion attenuation and the economic loss model. The results may at first glance seem bewildering through the many comparisons. However, we easily recognize a trend in that there is a growing concordance between the median results as the number of branches increase. The encouraging result is that with 27 or 81 branches in the logic tree calculations there are reasonable small variations in the obtained results. Finally Fig. 7 shows the computation time as function of number of branches on a relatively new laptop computer. While it is certainly desirable to include as many branches as possible in order to have an optimal grip on the confidence levels of the results, a short computing time is necessary in order to comply with the needs for rapid information. Fig. 7 indicates some 400 seconds of computing time for 27 branches. Considering the availability of increasingly faster computers it is likely that this computing time can be significantly reduced. Usually, when a real time damage scenario is going to be computed the researcher can assume, in a first approach, that the most uncertain parameters are those related with the earthquake source, the soil conditions and the attenuation relationship needed to estimate the ground motion that it will affect the building inventory. Then, 3 branches/options for each one of the these parameterizations will result in an initial logic tree with 27 branches.

Time (s)

10000

1000

100 Seconds

10

1 0 50 100 150 200 Number of Branches

Fig. 7. Computation time as function of number of branches in the logic tree. The computations were conducted on a Dell Inspiron 9400 computer with 2Gb of internal memory.

Results The results obtained above indicate that with a limited number of branches (due to the constraints on computation time) the SELENA software can be used in a near real time mode to evaluate damages inflicted by an earthquake. This conclusion is based on some conditions: • The preliminary results above indicate that the number of branches in the computations provide reasonably stable results already at around three branches per parameter. • The number of branches in the logic tree computations should not significantly exceed 27 branches unless a very fast computer is used.

333666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

In addition to the above basic conditions it is necessary to emphasize that more branches in the logic tree provides more sound confidence intervals for the final damage estimates than fewer branches. This is a classical trade‐off situation. While these results are promising we emphasize that more testing with different scenarios need to be conducted before any near‐real‐time implementation of SELENA can be done.

REFERENCES: 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”. Proc. of 12th European Conference on Earthquake Engineering, 696, pp. 1–10. 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. (2004), “Earthquake Risk Mitigation in Istanbul”. Chapter 7. In: Mulargia, F. and Geller, R. J. (Editors), Earthquake Science and Seismic Risk Reduction. Kluwer. Fajfar P. (1999). “Capacity spectrum method based on inelastic demand spectra”. Earthq. Eng. Struct. Dyn., 28, 979–993. FEMA (1999), “HAZUS 99. Earthquake Loss Estimation Methodology. Technical Manual”. Federal Emergency Management Agency, Washington D.C., USA. FEMA (2004), “HAZUS‐MH. Multi‐hazard Loss Estimation Methodology, Earthquake Model, Advance Engineering Building Module”. Federal Emergency Management Agency, USA. 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. 18–21 June 2006, Lillehammer, Norway. Ordaz, M., Miranda, E., Reinoso, E. and Pérez‐Rocha, L.E. (2000), “Seismic loss estimation model for México city”. Proc. of the 12th World Conference of Earthquake Engineering, Paper No. 1902. Wald, D.W, Quitoriano, V., Heaton, T.H., Kanamori, H., Scrivner, C.W., and Worden, C.B. (1999), “TriNet ʺShakeMapsʺ: 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”. Proc. of the Second International ISCRAM Conference (Eds. B. Carle and B. Van de Walle), 297– 299, Brussels, Belgium. Wyss, M. (2006), “The Kashmir M7.6 shock of 8 October 2005 calibrates estimates of losses in future Himalayan earthquakes”. Proc. of the Third International ISCRAM Conference, (Eds. B. Van de Walle and M. Turoff,), Newark, NJ (USA) (in press).

333777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

AAA 222‐‐‐DDD SSSEEEIIISSSMMMIIICCC SSSIIIGGGNNNAAALLL DDDEEETTTEEECCCTTTOOORRR FFFOOORRR SSSTTTAAANNNDDD AAALLLOOONNNEEE 333‐‐‐CCCOOOMMMPPPOOONNNEEENNNTTT SSSTTTAAATTTIIIOOONNNSSS

Matveeva T.1, Fedorenko Yu.V.2, Husebye E.S.3 1Institute of Geology, KarSC RAS, Petrozavodsk, Russia, [email protected] 2Polar Geophysical Institute, KSC RAS, Apatity, Russia, [email protected] 3Bergen Center for Computational Science, UNIFOB/UoBergen, Bergen, Norway, [email protected]

1. Introduction. Signal detection research become popular among scientists with the introduction of digital recording in seismology around 1970 not at least as this could relieve the analyst of tedious work tasks. An early design here was the so‐called STA/LTA‐detector where STA and LTA are short and long term noise and signal power averages. We term it a 1‐ D detector since only Z‐components recordings are used. Despite much ingenuous research still the STA/LTA‐detector is widely popular since a simple and robust alternative has not been found so far. The reason for this may be two‐fold; i) noise is not Gaussian so an optimum signal detector in a statistical sense will be hard to find and ii) close to the signal acceptance threshold the false alarm rate increases nearly exponentially. In other words, for say one hundred detections we have to reject 99 as signal look‐alike noise wavelets ‐ a truly hopeless task. Our alternative is to introduce a 2‐D STA/LTA‐detector; 2‐D in the sense that we utilize both vertical and horizontal record components jointly as expressed in eq. (1) below. In many signal detector studies much attention is often given to filter settings for ambient noise suppressions and STA‐length in time or samples. The efficient LTA‐length often exceed 1 min in its recursive definition. Below we first give principal detector design criteria before presenting our novel 2‐D detector extensively tested on the Cossack Ranger stations deployed in several countries.

2. Background. In previous works (e.g. see Fedorenko et al., 1998, Fedorenko et al., 1999) we demonstrated that the wavelet transform may be a more suitable tool than the FFT for signal detection, phase pickings, and signal source recognition. This in turn requires that the ambient noise is white which is equivalent to an approximate flat seismometer acceleration response curve. Since more instruments in use measures ground velocities the noise spectra are in general non‐flat as shown in Fig. 1 for the ARCESS array (N. Norway) center seismometer. To overcome this drawback in the context of signal detection the preamplifier of the Cossack Ranger seismograph is designed in such a manner as to modify our ground velocity measuring geophones to produce seismometer ground acceleration motions which in case of noise give approximately white spectra in the frequency range 1.5–20 Hz. The spectral hump at lower frequencies in Fig. 1, due to relatively strong microseisms, are in practice in no consequences as this part is removed by bandpass filtering in the detector processor.

333888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 1. Normalized power spectral density for ARCESS 3C center station (winter time) and our novel station at ASK (summer time) – in both cases Z‐ components. Since the noise level in winter is high the normalization gives low level of the high frequency noise at ARCESS. Note, that above 2 Hz the ARCESS spectra follows approximately 1/f slope while ASK spectra is almost flat in this frequency range.

In our more theoretical detector studies (Fedorenko and Husebye, 1999) we found that the popular STA/LTA detector has an excellent performance in comparison to many other detectors including the non‐ parametric Kolmogorov‐Smirnov (KS) one. It was one exception here namely in case of spiky records often caused by the electric outgauges in buildings where a station may be housed. In such cases the STA/LTA detector become literally blinded while the KS‐detector was little affected. Instead safeguarding here in terms of 2 different detector systems we build‐in a spike ʹkillerʹ in the A/D‐converter ensuring spike‐free records prior to signal detection per se. Another basic requirement in our 2‐D detector design is noise independency between components and the validity of this assumption is demonstrated in Fig. 2. As can be seen, there is no significant noise coherency even at low microseism frequencies (f < 0.5 Hz). In a typical signal detector passbands within 2–15 Hz the noise coherency is truly small being less than 0.02 units. Besides component independency also valid for components envelopes the noise appears to be Gaussian as presumed in our theoretical detector study. The above results imply that a 2‐D detector outperform the conventional 1‐D STA/LTA detector foremost because more signal information is incorporated in the test statistics.

3. The 2‐D signal detector. Our 2‐D signal detector operates in the three frequency bands which coincide with the corresponding frequency bands in wavelet transforms, namely 1.5625–3.125 Hz, 3.125–6.25 Hz and 6.25–12.5 Hz. The first step is data preparation prior to the detection process itself using IIR (Infinite Impulse Response) Butterworth filters of order 6 to

obtain filtered time histories for the xtkkk( )(N), yt ( )(E), zt ( )(Z) for our 3 frequency bands k =1, 2, 3 . The 2‐D detector is modeled after the 1‐D STA/LTA detector which in the former case is defined as: 221/2 STAHkSTAkk() t=+ IIRF ( x () t y ()) t

STAVkSTAk (tzt )= IIRF ( ( ) )

where H = horizontal and V = vertical, IIRFSTA defines Bessel low pass IIR filter of order 3 with cut‐off at 0.5 Hz. The LTAH and LTAV are defined in a similar manner but with a frequency cut‐off at 1/300 sec. The combination of Butterworth prefiltering and Bessel type of filters for forming the STA and LTA test statistics are motivated by the need of suppressing side‐lobe detections that is false alarms caused by noise triggering (Steinert et al., 1975).

333999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 2. Observed noise coherency between 3‐component waveforms and STAH and STAV envelops. This is original noise waveforms recorded by our novel ASK station of length 60 min presumably free of signals. Coherency at levels of 0.01 or less give component independency for the ambient noise field. Also note that similar results were obtained when noise waveforms were replaced by the corresponding envelopes, in this case the coherency is less than 0.001. In tha STA/LTA 2‐D detector the test statistics are the noise and signal envelopes.

Anyway, we may consider STAHk (t )

and STAVk (t ) as approximate envelopes

of horizontal ath () and vertical acceleration components in the given

frequency bands while LTAHk (t ) and

LTAVk (t ) are estimates of RMS of the ambient noise. Also, this LTA defenition is coincident with the parameter used for record ʹʹdenoising” in our wavelet processing scheme for picking automatically P‐ and S‐phase onsets (Husebye and Fedorenko, 1999). In order to to keep them from being affected by the seismic signal we freeze this values while the detector is in the

detection state. The 2‐D detector is taken to be ratio Rk ()t between the STA and LTA that is:

1/2 ⎡⎤22 ⎣⎦STAHkk (tt )+ STAV ( ) Rtk ()= 1/2 ⎡⎤22 ⎣⎦LTAHkk (tt )+ LTAV ( )

A signal detection is declared whenever any Rk ()t exceeds a preset threshold currently set at 4.0 for all three bands. This setting is rather conservative since we so far have not observed any clear false alarms. Depending on signal shape and duration the detector may be triggered several times during signal ʹpassageʹ. An independent trig requires a time lag of minimum 2 sec while detection state is defined as a time lag less than 60 sec between the last trigger events. We naturally count no of triggs during detection state as this parameter is useful for differentianting between earthquakes and explosions (most 8 triggs) while man made signals like marine air gun activities, human walking near the site etc. often produce more than 20 triggs. The 2‐D detector parameterization may be tailormade for each individual site reflecting local knowledge on ambient noise and signals triggered by human activities. The horizontal and vertical acceleration envelopes are used to obtain the estimate of an env acceleration envelope atk () and of an apparent incidence angle θk 1/2 env ⎡⎤22 atkkk()=+⎣⎦ STAH() t STAV() t ⎡⎤env θkkk(ttat )= arcsin⎣⎦ STAH ( ) / ( )

Seismic events have 10<θk ()t <60 while spikes and non‐seismic events often fall outside this angular interval.

444000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 3. Illustration of 2‐D detector performance. Probability density function pR(,)θ modeled for uncorrelated 3‐component Gaussian noise with zero mean and same variance in each component. Critical region for the 2‐D detector appear far from “p.d.f. main body” assuring small false alarm rate. Insert shows one‐dimensional p.d.f of R and of conventional STA/LTA ratio Sz =STAZ/LTAZ. It is clear that R statistics produce less false alarms for uncorrelated noise because its variance is less than variance of Sz statistics by a factor of 3 approximately.

4. Results – 2‐D detector performances. Detector performances are not easy to quantify since close to the threshold value it is difficult to validate the event origin – earthquake or false alarms. The problem is severe as ca 40‐50 percent of reported event detections for an array or a station remain unassociated meaning that the potential event has no counterpart in station logs elsewhere. In Fig. 3 we have simulated the 2‐D signal performances for uncorrelated 3‐ component Gaussian noise field. Clearly, we expect the 2‐D signal detector to be superior to its 1‐D counterpart by a factor of 3 approximately. In another experiment a CR‐II station and a Kinemetrix Ranger station shared the siting at Ask west of Bergen (Norway). The 2‐D detector of the CR‐II station proved superior to that of the adjacent 1‐D detector station. Signal detector parameters may be utilized for more that just detecting probable seismic event recordings. For example, Mendi and Husebye (1994) demonstrated that the STA parameter would provide a magnitude estimate as well given an assumption about likely epicenter distance. In case of STA measures reflecting pure noise estimates the corresponding magnitude estimates give a lower limit of the stations event detectability. The detection parameter ʺapparent angle of incidenceʺ is observationally limited to the observation interval 10oo− 60 for earthquakes and chemical explosions (Fig. 3). This implies that horizontal components record parts of the P‐wave field even for teleseismic events making 2‐D detectors superior to those based on vertical components only. For arrays where vertical sensors are dominant (only four 3‐comp. stations in ARCESS) 2‐D detectors are of less interest unless the later ones are used for detecting Lg an Rg‐waves. The conservative threshold setting of 4.0 ensure no false alarms due to noise triggering. To lower threshold value, for Gaussian noise a threshold of 2.4 would be acceptable, we would produce many more detections of air gun signals and nearby human activity signals. Such events are of no scientific interest so better removed in this manner than by conservative threshold setting.

444111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

The novel 2‐D signal detector has proved superior to the conventional 1‐D STA/LTA for stand‐alone 3‐component stations. In a monitoring context, we want to extract more information on detected signals per se and this would be feasible using continuous sampled STA parameter for representing the original waveforms. The STA sampling rate is often 1 Hz while the original waveform may be 50 or 100 Hz so even in under adverse field conditions waveform transfer to the HUB would be affordable say using mobile phones (Katkalov and Husebye, 2007). Two of us, Fedorenko and Husebye, have previously demonstrated that envelope waveform can be used for highly accurate epicenter locations and source type identification (Fedorenko et al., 1998, Husebye et al., 2002).

Summary remarks. For efficient monitoring of local earthquake activity a network of stand‐ alone 3‐component stations would be most efficient in comparison to a few small arrays and much more cost efficient. A smart signal detector is an essential element in such an undertaking and the novel 2‐D detector presented here is well suited for such tasks. Its performances are better than similar 1‐D detectors both in terms of lower signal detection threshold and false alarms rates. In addition, detector outputs in form of envelope traces can be used for desired monitoring results in terms of epicenter locations and source identification.

REFERENCES:

Fedorenko, Yu. V., E. S. Husebye, B. Heincke, and B. O. Ruud, 1998. Recognizing Explosion Sites without Seismogram Readings: Neural Network Analysis of Envelope‐Transformed Multistation SP Recordings 3‐6 Hz, Geophys. J. Int., 133, F1–F6. Fedorenko, Yu. V., E. S. Husebye and B. O. Ruud, 1999. Explosion Site Recognition: Neural Net Discriminator Using Three‐Component Stations, Phys. Earth Planet. Int., 113, 131–142. Fedorenko, Yu. V. and E. S. Husebye, 1999. First Breaks – Automatic Phase Picking of P‐ and S onsets in Seismic Records. Geophys. Res. Lett., 26, 3249–3253. Husebye, E. S., Fedorenko, Yu.V., Beketova, E.B., 2002. Enhanced CTBT Monitoring through Modeling, Processing and Extraction of Secondary Phase Information at High Signal Frequencies. 24th Seismic Research Review – Nuclear Explosion Monitoring: Innovation and Integration, Sep. 17– 19, Ponte Vedra Beach, FL., p. 292–301. Katkalov, Yu., Husebye, E.S., 2007. Cosack Rager field opration – efficient data transfer via mobile phones. Manuscript in preparation. Mendi, C.D., Husebye, E.S., 1994. Near real time estimation of magnitudes and moments for local seismic events. Ann. di Geofisica, Vol. XXXVII, 365–382. Steinert, O., E. S. Husebye, H. Gjoystdal, 1975. Noise Variance Fluctuations and Earthquake Detectability. J. Geophys. 41, 289‐302.

444222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

TTTHHHEEE KKKAAARRREEELLLIIIAAANNN RRREEEGGGIIIOOONNNAAALLL SSSEEEIIISSSMMMIIICCC NNNEEETTTWWOOORRRKKK ––– TTTHHHEEE CCCOOOSSSSSSAAACCCKKK RRRAAANNNGGGEEERRR IIIIII SSSEEEIIISSSMMMOOOGGGRRRAAAPPPHHH

Matveeva T.1, Fedorenko YU. V.2 and Husebye E. S.3 1Institute of Geology, KarSC RAS, Petrozavodsk, Russia, [email protected] 2Polar Geophysical Institute, KSC RAS, Apatity, Russia, [email protected] 3Bergen Center for Computational Science, UNIFOB/UoBergen, [email protected]

1. Introduction. Present days when seismic data acquisition system and network operational cost are becoming affordable, earth scientists pay more attention to the study of tectonic activities of low‐seismicity regions. The conditions for seismic monitoring of such regions are dense seismic networks which can guarantee weak event detection down to a ML ~ 1.5. This was the challenge when starting the deployment of the Karelian (NW Russia) seismic network namely the necessity to have an affordable 3‐component seismograph, capable of providing adequate seismic records for basic research. An additional requirement is ease of seismograph installation and maintenance at remote sites. Considering many seismograph makes, mainly these, based on inexpensive geophone sensor, we found not surprisingly that the Cossack Ranger II seismograph (CR‐II) best meet the above requirements. This instrument initially developed by Fedorenko in 1999 has been operated successfully in the most adverse climatic conditions like Kirovsk (Chernouss et al., 1999), W. Norway (Fedorenko et al., 2000) and also in the tropical Costa Rica (Waldo Taylor, pers. communication).

2. CR II – General Description. The Cossack Ranger II 3‐component seismic station is an advanced, inexpensive and low‐power 3–12 channel seismograph. It consists of four modules being common for most seismic station designs i) 3‐component seismic sensor unit with or without preamplifier, ii) an A/D‐converter unit, iii) optional GPS clock and iv) data logger based on any type of computer (desktop, notebook, Palm Digital Assistant etc.) supporting USB or RS232 port. CR‐II can operate as a 3‐component station or 9‐element mini‐array storing data locally or remotely using TCP/IP protocol via Ethernet, GPRS modem, radio transmitter, wireless 802.11b etc. The CR‐II data acquisition software supports continuous, level‐triggered and STA/LTA triggered recordings. It can also be easily configured to start and stop data acquisition in given time intervals. While operating in a triggering mode the loads on data links are considerably reduced, and even slow modem connections will be adequate. To synchronize station time to UTC the CR‐II timing system uses either the GARMIN GPS 18 clock or Internet Time Protocol service via TCP/IP connection. Recorded data can be stored locally to hard disk or sent to the central Hub in a few minutes after event recording ʹterminatedʹ. Optionally data may be written to a flash memory card which can be exchanged within a few minutes. Adequate power supply protects battery from over‐discharging and easy switching to exchange batteries. A standard housing conforming to the IP68 standard is used to protect equipment from dust and liquids.

3. Field installation of the Cossack Ranger II seismograph. Any seismograph installation requires i) power access, ii) timing device or clock, iii) remote data transfer and iv) proper siting rock. Since the CR initially was designed for school yard installation we would address the above requirements in this context. First electric power in terms of the 220V electric

444333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

network or 9V batteries. The former is most convenient and our preference is given for the automatic rebooting at power out ganges say due to lightening. Alternatively, rebooting is performed remotely from the Hub or network center. For remote, stand alone installations batteries provide adequate power supply but is often somewhat inconvenient with frequent replacement unless consumption of ADC and data logger is exceptionally low. The CR‐II design is flexible permitting stand alone operations with power supply from batteries. Accurate, absolute timing is a necessity for comparing and jointly analyze recordings from 2 or more stations. Before the advent of the GPS timing device this was often a severe practical problems. Clocks, even expensive ones, had a timing accuracy seldom better then 0.1 sec. For field installations, similar accuracy was attainable, using radio clock signals from Potsdam (Germany) and/or Rugby (UK) for calibrations. When we first deployed our CR‐ stations in W. Norway high schools in 2000/2001 we used the Internet timing system being accurate to +/‐ 0.01 sec or better. At that time GPSs were relatively expensive and besides a drawback is its required outdoor monitoring with exposures to rainy weather and theft. Present day, the preference is for using highly accurate and inexpensive GPS timing (+/‐ 0.002 sec). For most monitoring systems today fast data access at the local center or Hub is mandatory. This has been feasible since the 1970‐ties using say leased commercial telephone lines but high costs have prevented wide spread use of such solutions until recently. Today the relatively easy access to the Internet at almost no costs make this mean for the data transfer universally praised by seismologists. In case of stand alone station operations most of the events recorded and stored on disks in the field while only an affordable fraction of the data stream is transferred within a reasonable time lag to the Hub. In this context is interesting to ʹconvertʹ mobile phones into a data transfer unit (Katkalov and Husebye, 2007). Note, advanced data logger design permit extractions of critical parameters for an epicenter location like P – arrival times and amplitudes and also signal envelopes (1 sec sampling) tied to the Hilbert transform of the original waveforms (100–200 Hz sampling). Proper seismograph siting is critical for the high quality network operations. Preferably site surveys proceed the actual installation measuring the ambient noise field and avoid locations close to fast running machinery. In particular, it is important to avoid any site ʹtied toʹ parts of the housing structure which easily ʹtransmitʹ in‐house activities. Switching on‐ and off‐ of machinery are occasionally manifested as spikes in the recordings. If the spikes are simple and of a regular form they are easily removed by a digital filter, the CR‐II data logger incooperates such a feature. 4. Karelian seismic network – CR‐II installation. In Karelia we undertook a special study of CR siting environments by installing 4 CR stations at very different sites (Figure 1).

444444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 1. Seismic stations: PTRZE, KOSTE, BELME and PITKE (triangles) with the CR‐II instrumentation installed in 2000–2006 in Karelia (NW Russia) and adjacent quarries (red stars).

The network installation process began in Aug. 2000 when the first station was installed in the Geophysical Yard of the Petrozavodsk University (PTRZE in Fig. 1). Seismometer was installed in the basement of a wooden, two‐story house on solid rock. Data were transmitted via radio link to the Institute server. This station has provided reliable event recordings for more than 6 years. Then during summer 2006 a new station was installed in a school yard (KOSTE in Fig. 1). It was impossible to dig a hole close to the data logger and due to a lack of safety for outdoor installation we put sensor on the floor in the school building. As expected, this place was hopeless due to the human activity in the building so only explosions from the nearby mine were recorded during the half year of operation. The remaining two stations (BELME and PITKE in Fig. 1) were installed on the premise of radio masts where service buildings provided stations with proper housing. Therefore one station BELME was installed near the mast in the outskirts of the small town Belomorsk close to the White Sea. The sensor was installed in an abandoned building on exposed rock. The closeness of the town, railway station and White Sea bay made reliable recording impossible of even close occurring events. Only a few teleseismic earthquakes and some explosions were recorded by this station over 6 months. The other station PITKE was installed near a mast relatively far from any human activity and 3 km away from the Ladoga lake. Sensor was installed on the solid rock and a wooden shielding was made. This installation turned out to be successful and events of different kind were recorded. For both of these stations (BELME and PITKE) data transmission and station operation were organized via GPRS modem of mobile phones. An unexpected shortcoming of the GPRS connection was during lunch and dinner time when the cellular network was overloaded causing low speed connection. Contrary to our expectations the electric lines of the radio masts did not produce spikes on the recordings and likewise wind vibrations of the masts had no significant recording influence.

5. Cossack Ranger Recordings in Research. The first Cossack Ranger seismographs were to some extent home made in the sense that besides the individual sensor of cost $ 60 each, partly ʹhome madeʹ ADC and an old PC as data logger had a price tag less than $ 1000. Some colleagues asked whether such a dirt cheap seismograph was useful for advanced research purposed. This was considered a challenge so the first step in testing recording quality was to site the CR adjacent to a Kinemetrix Ranger seismograph – the ASK station close to the Institute and Bergen city itself. Test was only partly successful as the Kinemetrix recordings obviously were clipped for a local ML~3.5 events. Further tests were a polarization study of Karelian events (Matveeva et al. ibid) demonstrating that observed and calculate wavefield

444555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

polarization properties were similar. This would not have been feasible with anomalous phase shift among the sensor components. In another study, we used deformable templates technique for linearly projecting the Lg‐wave train from one CR station into another one – the original waveforms replaced by their respective Hilbert transforms. In other context, we subjected the Pg/Pn‐coda to an automatic phase picking scheme and then using Bayesian statistics for validating the pickings toward updated focal parameter estimates including depth. The CR is also efficient in monitoring local seismicity since its 2D signal detector is superior to the conventional 1D detector commonly used.

Summary. All installed CR‐II stations proved itself as robust, stable and easy to install (even for a single person) and operate seismograph. Some of the stations operated for more than a year in Norway and Karelia without any disruption while in some cases water leaked into the instrumentation boxes. This problem was eliminated by using more elaborate box design and also to ensure that the sealing putty in fact stopped water sipping inside the box. We have also operated the CR’s in most moistly environments like the jungle in Costa Rica, the wet west coast of Norway and cold areas like Karelia, Spitsbergen and in the Kirovsk mountains. It has been functioning equally well in all these environments. To summarize, the Cossack Ranger design is well suited for advanced seismological research and network monitoring of local seismicity.

REFRENCES:

Chernouss, P., Zuzin, Yu., Mokrov, E., Fedorenko, Yu.V., Kalabin, G., Husebye, E.S., 1999. Avalanche Hazards in Khibiny Massif, KOLA, and the new Nansen Seismograph Station. IRIS Newsletter, No. 1, p. 12–13. Fedorenko, Yu.V., Husebye, E.S., Bulaenko, E., 2000. School Yard Seismology. Orfeus Newsletter, Vol. 2, No. 3, p. 22. Katkalov, Yu., Husebye, E.S., 2007. Cossack Ranger field operation ‐ efficient data transfer via mobile phones. Manuscript in preparation.

444666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

AAAUUUTTTOOOMMMAAATTTIIICCC PPP‐‐‐CCCOOODDDAAA PPPHHHAAASSSEEESSS IIIDDDEEENNNTTTIIIFFFIIICCCAATTTIIIOOONNN UUUSSSIIINNNGGG BBBAAAYYYEEESSSIIIAAANNN AAAPPPPPPRRROOOAAACCCHHH

Matveeva T.1, Fedorenko Yu. V.2, Fedorenko M.3, Husebye E. S.4 1Institute of Geology, KarSC RAS, Petrozavodsk, Russia, [email protected] 2Polar Geophysical Institute, KSC RAS, Apatity, Russia, [email protected] 3Kola Branch of the State University of Petrozavodsk, Apatity, Russia 4Bergen Center for Computational Science, UNIFOB/UoBergen, Thormoehlensgt. 55, 2008 Bergen, Norway, [email protected]

1. Introduction. Reliable focal depth estimation could be provided with confidentially using secondary phases in earthquake recordings. Standard analyst practice is to pick manually and identify a ʹphaseʹ in the P‐coda in terms of beliefs. These efforts are often useless because phase pickings are not coupled to a phase validation scheme. With the latter it is meant that a phase picking, say a pP‐phase candidate, should comply with preceding P‐polarization and P‐ slowness requirements. Obviously, picking secondary phases in the P‐wave coda is not only a technical problem but foremost a scientific one; how many phase candidates may be picked and what record screening procedures should be used? Extracting a large number of phase candidates from a network of stations and then introducing a maximum likelihood method for recognizing them properly and jointly with a focal depth estimate requiring excessive computations (Husebye et al., 2002). So at this stage we limit ourselves to a phase validation scheme based on signal polarization characteristics. Next we outline a Bayesian approach for screening depth phases arriving in the P‐wave coda between Pn/Pg and Sg/Lg. Note, S‐waves propagate in 2 modes ‐ SV and SH and are in generally relatively complex and may not be well polarized due to small unknown ray path differences. Avoiding such complications we leave S phases validation for future investigations.

2. The phase picking scheme. Most applications of our novel concept are at local and regional distance ranges because wave amplitudes on the horizontal components are relatively week at teleseismic distances. Now we focus on validating the classical depth phases pP, sP and PmP which are needed for accurate local depth estimates of crustal earthquakes. Starting with a rough epicenter location we have apriori information on expected direction (azimuth) of particle motion projected in the horizontal plane. The simple two‐layer velocity model predicts propagation time reasonably well, but fails to predict the value of the incidence angle apparently due to uncertain velocity gradient near the surface. Although we do not know the incidence angle per se, we believe that Pn, pP and sP phases arrive with smaller incidence angle than Pg. We also believe that seismic phases are better polarized than random coda and noise wavelets and that the probability to pick and identify correctly seismic phases (pP, sP etc.) increases with their amplitudes. In this study we attempt to transform our degrees of belief into numbers incorporating the information contained in waveforms. As a first step we consider the “wrong” phase screening problem as clustering and simply want to verify the existence of some clusters (or classes) using signal polarization properties. Furthermore, we try to answer if the discrimination between classes are good enough and how those classes correspond to our physical understanding of wave propagation. As the second step we will apply accumulated experience to relate clusters to seismic phases.

444777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

We assume that any P‐wave onset is a plain sum of linearly polarized signal wavelets and scattered non‐polarized wavelets and noise. The p.d.f. of seismic signals is quite close to Gaussian so we presume that the covariance matrix holds almost all information contained in the waveforms. The preparation of input for our clustering analysis consists of the following procedures:

1. The 3 – component waveforms s()tk , tktk = Δ , k = 01,...,K − are filtered to suppress background noise and if necessary, the high frequency part of signal spectra where the polarization usually degrades;

2. The real signal is converted to analytic signal sˆ k = s k + jH{}sk where s k = s()tk , j =−1 and H{}. denotes the Hilbert transform operation; T* 3. The covariation matrix Csskk= ˆˆ()k is calculated for each k , then each element of Ck is averaged over time by a low‐pass linear phase Bessel type IIR filter or Savitsky‐Golay filter. To reduce the computational burden in forthcoming operations these elements are decimated (usually by factor M = 5 .. 10, according to the smoothing filter used);

4. The eigenvalues Λ k and eigenvector u1k are evaluated for each sample k for smoothed

and decimated version of the C% k . Then we calculate u1k , v1k , λ1k and rk with k = 01,..,KM / − . Again the parameters used in analysis are the incidence angle observed and expected, azimuth and arrival times. The last two parameters stem from the preliminary estimate of the epicenter location.

3. Classification. Classification problems are discussed in many books and articles so here we outline the appropriate one for our analysis (Merhav and Ephraim, 1991; Duda et. al, 2000; Tarantola, 2005). One meaning of “classification” is that our task is to establish the existence of classes or clusters in the data. The other meaning is to find a rule that will enable us to classify a new set of data into one of existing classes. The former one is known as the Unsupervised Learning or the Clustering problem, the latter as the Supervised learning. Below we consider the Clustering problem, that is i) to establish clusters in seismic signals and ii) relate them to seismic phases if feasible. There are three essential components of our combined approach: 1. Estimate prior probability distribution (the relative frequency with which the respective classes occur in the population); 2. Establish rules for separating classes; 3. Set the penalty associated with making a wrong classification. Applying Bayes theorem we can calculate the aposterior probability for all established

classes Ai Nφ −1 p()Aiiiii|=xx L( | ApA )() /∑ L ( x | ApA )() and assign result of measurements x to the i=0

class Ad with maximal aposterior probability given x . In our case all variables are real numbers so probabilities become probability density functions (p.d.f.) and aposterior

probability related to Ai must be obtained by integration over model space.

The discriminator is the expected incidence angle φE which is not exactly known. The

uncertainty of our knowledge on φE within each class is described by a truncated Gaussian p.d.f. Also we assume that the deviation from the estimated azimuth of P‐phase arrivals does not differ much for all P‐phases because the deviations of the crust velocity model from horizontally layering are not large. Preliminary location is approximate therefore we employ

444888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

our priory knowledge on azimuth in terms of a probability density function which we also define as truncated Gaussian p.d.f. Finally we construct the p.d.f. of model as −−φφ2 /2 σ2 2 2 ′ ()Ei φ −−()θ θσE /2 θ fAki()=,,,,,∝ f (φθ λ1max1 r| φ EiE θ λ ) r kk λ e e . Simplifying our p.d.f. we get likelihood for each waveform sample k which is used to separate classes: ⎡⎤⎡⎤22 ()θθdd−−() φφ luki()1D,∝−v 1 |exp A ⎢⎥⎢⎥ ⋅− exp . We choose this form of the likelihood 22ss22 ⎣⎦⎣⎦⎢⎥⎢⎥θφ because we can not avoid numerical integration over θ . Calculating and plotting posterior

probabilities for each Ai versus time, also assuming that our distributions are narrow we can safely expand our integration limits to ± infinity and obtain a closed form solution: 22 ⎡⎤⎛⎞22 ⎛⎞ 22 ⎜⎟σφdiE++ss φ2222 ⎜⎟ σθ dE θ ⎢⎥⎝⎠φφφθddφθiE ⎝⎠ θθ E PrArik |x k∝−−+−− kλ1′ k exp . () 2222⎛⎞22 2222 ⎛⎞ 22 ⎢⎥22σσss⎜⎟++22ssσ σσ ss ⎜⎟ 22σ ⎣⎦φφ⎝⎠ φ φφ φθθ θθ ⎝⎠ θ θ Evaluation of the cost for wrong classification is itself a quite difficult problem. It is coupled with the target problem of hypocenter location by maximum likelihood because the wrong phase classification affects the reliability of the final solution in an as yet unknown way. At this stage of development we do not elaborate on any cost thus in effect assuming that all costs are equal.

4. Testing Results of Polarization Concept for Secondary Phases Picking. To show here the viability of our novel polarisation concept for picking depth phase candidates we have analyzed the local earthquake recording from the Petrozavodsk (PTRZ) station in Karelia, NW Russia. This station is equipped with the Cossack Ranger 3‐component seismographs (Fedorenko et al., 2000). Our polarization concept for phase picking and validation appear to work satisfactorily (see Fig. 1). In estimating phase arrival times we used the Karelian crustal model with Moho at a depth of 42 km and the corresponding Pg, Pn, Sg and Sn; phase velocities are 6.2, 8.0, 3.8 and 4.7 km/sec respectively (Sharov, 2004). For this event we presumably detected Pn, sP, and Pg but no pP. The latter may reflect a particular orientation of the fault plane with no pP focusing towards the PTRZ station (Pearce, 1996) while PmP is expected to be weak at epicentral distances exceeding 500 km. A puzzling feature is the relative steep Pg incident angle at 40 deg while that of Pn is 46 deg. There is as mentioned no rational explanation for this except that crustal wave propagation is often complex for nearly horizontally propagating waves. Figure 1. Secondary phases automatically picked by our polarization analysis schemes. The recording at the Petrozavodsk seismic station (PTRZ) from a small local earthquake on 06 December 2002 at an azimuth angle of 344 deg and distance 536 km. Upper plot shows

Pr(φEik| x ) for each data sample xk . Middle plot shows the Z‐component trace. Lower plot shows the result of a polarization filter (solid line) and shift in azimuth angle for particle oscillations (dotted line). For detected secondary phases the shift in azimuth angle equals to zero and the filter output reaches the maximum. Arrival times calculated from given location relative to the Pn onset are: sP +6 sec, Pg +9.6 sec.

444999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

5. Discussion. In this work we attempt to use P‐signal characteristics for isolating secondary phases of the P‐type. Reliable identification of pP‐ and sP‐phases are critical for accurate focal depth estimates as conventional methods like cepstrum analysis or subjective analysts picks are not efficient (Bonner et al., 2002). In testing performances of our polarization scheme on recordings from school networks (W. Norway) and Karelian Seismograph Network (NW Russia) we obtained gratifying preliminary results. Firstly, arrivals outside the ‘true’ azimuth range of say ± 5 degrees are rejected so interferences from side‐reflections are small. Likewise incident angles tied to crustal/upper mantle velocity models are also efficient in constraining false alarms that is interferences of scattering wavelets. Even with a somewhat limited analysis experience we are confident that the polarization scheme works satisfactorily. Still some judgement in interpreting results may be needed and the most obvious case is as follow: the probability function plot indicates a phase arrival but the trace itself exhibit only a blurred phase arrival and beside no significant change in azimuth angle in the polarization filter trace plot. So far we have not succeeded in finding so‐called Ground Truth events with focal parameters accurate to within 3 km. Then we would have an adequate basis for judging performances and whether on a routine basis can extract reliable sP‐, pP‐ and PmP‐phase arrivals. If so we will have reliable focal depth estimates also for weak local and regional earthquakes – this has been an outstanding research goal for several decades.

REFERENCES:

Duda, R.O., Hart, P.E., Stork, D.G., 2000. Pattern Classification. Wiley & Sons, 680 p. Fedorenko, Yu.V., Husebye, E.S., Boulaenko, E., 2000. School Yard Seismology. Orfeus Newsletter, Vol. 2, No. 3, 22 pp. http://www.orfeus.nl/v2no3 Husebye, E.S., Fedorenko, Yu.V., Beketova, E.B., 2002. Enhanced CTBT Monitoring through Modeling, Processing and Extraction of Secondary Phase Information at High Signal Frequencies. 24th Seismic Research Review ‐ Nuclear Explosion Monitoring: Innovation and Integration, Sep. 17–19, Ponte Vedra Beach, FL., USA, 292–301. Merhav, N., Ephraim, Y., 1991. A Bayesian Classification Approach with Application to Speech Recognition. IEEE Transactions on Signal Processing, vol. 39, No. 10, Oct., 529–532. Sharov, N.V., (ed.), 2004. Deep structure and seismicity of the Karelian region and its margins. Karelian Research Center, Institute of Geology, Petrozavodsk, Russia, 352 p. Tarantola, A., 2005. Inverse problem theory and methods for model parameter estimation. SIAM, University City Science Center, Philadelphia, PA, USA, 358 p.

555000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMIIICCC RRREEESSSIIISSSTTTAAANNNCCCEEE OOOFFF SSSYYYSSSTTTEEEMMMSSS AAANNNDDD EEELLLEEEMMMEEENNNTTTSSS OOOFFF EEEXXXIIISSSTTTIIINNNGGG IIIGGGNNNAAALLLIIINNNAAA NNNPPPPPP OOONNN TTTHHHEEE BBBAAASSSIIISSS OOOFFF PPPEEERRRFFFOOORRRMMMEEEDDD IIINNNVVVEEESSSTTTIIIGGGAAATTTIIIOOONNNSSS

Mereznikov A. Ignalina Nuclear Power Plant, [email protected]

Abstract: Ignalina NPP was designed and constructed in accordance with standards and regulations that did not stipulate the consideration of seismic loads and performing of additional calculations. Ignalina NPP power units 1 and 2 are similar according to the design of construction, main systems and elements. Special investigations of ground motions on the site were performed in 1988, after the design and construction of the INPP both units has been finished. Probabilistic characteristics of floor response spectra of Ignalina NPP buildings under earthquakes are calculated in 1991. The maximal characteristics on earthquake intensity for blocks A, B, V, D building 101 – IMCE=6.5, maximal ground acceleration ~ 0.75 g (according to the data for power unit 1); pump station SWS 120/1.2 –‐ IMCE =6.0, maximal ground acceleration ~ 0.6 g, ECCS pressure tanks, building 117/1.2 – IMCE=7.0, maximal ground acceleration ~ 0.1 g were adopted for calculation of initial characteristics of Ignalina NPP constructions under MCE. The maximal intensity on Ignalina NPP site from distant earthquakes (only Carpathian and Skagerrak locations were taken into account) will not exceed 5 of MSK‐64 numbers. Number of seismic impact assessments was performed in 1994–1997 by Western experts. The checking calculations of the primary circuit critical elements, including DS and its supports, ECCS elements, electric power supply, RM of Ignalina NPP were performed in 1998– 2001 with additional participation of scientific and engineering support organisations which applied the modern computer codes of calculation and data of the actual structures of INPP. New objects, systems, elements of Ignalina NPP are designing and constructing/buying considering the modern standards of design. It is not reasonable, considering hypothetical seismic risk, to grade all the elements and systems of existing NPP to the state qualified as ʺseismically provedʺ according to the modern standards. After the earthquake corresponding according to magnitude to the design one or exceeding it, in accordance with Rules on the Design and Safe Operation of the Equipment and Pipelines of NPP the extraordinary monitoring of equipment and pipelines is stipulated. The performed investigations on assessment and improving of seismic stability of the systems and elements of Ignalina NPP show that the activity of INPP, as an Operator, in improving of seismic stability to the level of requirements of modern standards of safety, corresponds with the international practice, as it was stated in IAEA report INSAG‐8 [1]. It is the responsibility of the Operator of NPP to use the latest available information. According to the requirements of modern Lithuanian standards the designing and the construction of new nuclear facilities have to consider the seismic impacts [2, 3]. Ignalina NPP implements the system of seismic alarm and monitoring. Now this system is put into trial operation. Analysis of data obtained by SMS allows to estimate whether structures and equipment are damaged after earthquake and, consequently, whether plant is able to operate further. Information obtained by SMS about local earthquakes and

555111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

long‐distance seismic events is continuously transmitted to the Lithuanian Geologic Survey for analysis. The obtained data is used to study the seismicity of the Baltic Region and Lithuanian territory [4].

REFERENCES:

1. Safety analysis report for Unit 1, 2, INPP. 2. Program for categorisation of safety related building, structures, systems and their components under seismic evaluations. 3. Design Documentation for New Facilities under Decommissioning of Ignalina Unit 1, 2. 4. Operation Manual of Seismic Alarm and Monitoring System.

555222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMIIICCC HHHAAAZZZAAARRRDDD OOOFFF TTTEEERRRRRRIIITTTOOORRRYYY OOOFFF LLLOOOCCAAATTTIIIOOONNN OOOFFF TTTHHHEEE IIIGGGNNNAAALLLIIINNNAAA NNNUUUCCCLLLEEEAAARRR PPPOOOWWWEEERRR PPPLLLAAANNNTTT BBBYYY RRREEESSSUUULLLTTTSSS OOOFFF CCCOOOMMMPPPLLLEEEXXX RRREEESSSEEEAAARRRCCCHHHEEESSS 111999888777–––111999888888 OOOFFF TTTHHHEEE LLLAAASSSTTT CCCEEENNNTTUUURRRYYY

Mindel I.G., Trifonov B.A. Russian Academy of Sciences, Institute of Environmental Geosciences P.O. Box 145, 13‐2, Ulansky per., 101000, Moscow, Russia. E‐mail:[email protected].

In 1987‐1988 of last the century, after introduction in action of “Regulation on aseismic design of plants” PNAE G‐006‐87 the decision of Council of Ministers of the USSR formed the Governmental commission for consideration of some questions of safety of operation INPP, and also tool researches were carried out with the purpose of seismic microzonation into districts of a site of station. In tectonic aspect INPP site settles down on Zaralaisk step in area of interface of three large tectonic structures of the crystal base. The roof of the base lies on a depth of about 700 m. The site is in 10 km from East‐Lithuanian Zone of faults and in 30 km from Schilutsk–Polotsk zone of the faults, which are made active at the newest stage of development. Zone of the Daungavpils break is tracked in 15–20 km, linear active Aukschaits zone is traced in 40 km in a longitudinal direction. Units of crossing East Lithuanian and Zaraisk–Breslavsk, and also Daugavpilse and Reznaksk breaks are located approximately in 15 km to northeast from INPP. In October–November 1988, the subcommittee on a question of seismicity has been formed in structure of the Governmental commission on development of actions on increase of safety INPP. The largest experts in the field of seismicity and seismotectonic have been enlisted in this subcommittee: V.V. Schteinberg (chairman), N.V. Schebalin, G.I. Reisner, P.I. Suveizdis and others. The conclusion by results of work of this bureau has been considered and approved on bureau MSSS at Presidium of Academy of Sciences USSR 28.11.88. Results of estimation of maximal magnitude (Mmax) and intensity of earthquakes in an epicenter (IΔ) in area INPP are listed in Table 1 according to commission materials. Table 1 The minimal Intensity in Depth of the The name of zone Mmax distance to a epicenter I , center h, km Δ site, km unit The East‐Lithuanian zone limited in NNE to crossing with Daugavpilsk and Zaraisk–Breslavsk by a fault, in 4.5–4.6 10 10 5.7–5.9 the south‐crossing with Schilutsc– Polotsk and Aukschtaitsk faults Unit of crossing of the East– Lithuanian zone with Daugavpilsk 4.8 10 15 5.9 and Zaraisk‐Breslavsk faults Scattered seismicity in case of an adverse location of the center 4.5–4.6 10 0–50 6.2–6.4 directly under a site

The catalogue of earthquake with the intensity of 3–7 points in radius of 250 km from the site INPP contained data on 19 earthquakes for a period since 1616 till 1909. In result it was decided to accept IDBE=5 points and IUDBE=6 points for the soils II categories on seismic

555333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

properties. INPP site test also suffers shocks from remote deep focus earthquakes from Carpat zone and the Scandinavian centers such as Skaggerat earthquake of 1904. The maximal force of concussions from the centers of remote Carpat and Scandinavian earthquakes does not exceed the intensity 4–5 points. Engineering‐geological conditions of INPP territory are very complicated, that is caused by geological and litological structure and hydro‐geological conditions. The top part of Quaternary massif up to depth of 50 m is composed of end‐moraine by the formations of the Valdai glaciations presented by both layered and facial‐replaced moraine by loams and sandy loams with inclusion of gravel and pebbles and water‐glacial sand with inter layers of dusty and fine differences. In downturn of a relief there lie lake‐glacial deposits: loams, sandy loams and varved clay up to 7.6 m thick blocked by lake‐bog formations (peat, loams, sandy loams, sand, sapropels) of 0.5–10.9 m thick. In sites of power units bog deposits are replaced by artificial soils. All investigated soil massif within the limits of a site of the atomic power plant was watered. Depth level groundwater rather changed from 0,5 up to 8 m. Near to power units levels of ground water are a little bit reduced due to influence ring a drainage. On materials of engineering‐geological researches 11 engineering‐geological elements have been allocated, parameters of physical and physical‐mechanical properties of the soils were resulted and the category of the soils on seismic properties in each element was estimated. Practically, the soils of all 11 engineering‐geological elements according to table 1 SNiP II‐7‐81* [1] fit to III category on seismic properties. Tools researches consisted in surface seismic investigation (a method of the refracted waves) and borehole investigation (vertical seismic profiling), and also in registration of fluctuation of soils from various sources in points of a site typical from engineering‐geological conditions. Seismic researches was performed with the propose of determination of velocity shear (S) and longitudinal (P) waves in soils the top part of the sections for the subsequent estimation of increments seismicity on a method of comparison seismic regidity (MSR), and also for drawing up of the seismic section (models) further used in calculations of parameters of seismic actions. Special tool researches have been carried out for an estimation of absorbing characteristics of soils the top part of section at propagation of seismic waves. At registration of fluctuations of the soils in three points of a site within 2 months at rather high level of handicapes (work of turbogenerators, building mechanisms, transport, etc.) 12 events have been written down, amplitudes of fluctuations at which 3–4 times exceeded a background of handicapes. All records are not similar to fluctuation from earthquakes and are man‐caused by sources (sharp changes of loading turbogenerators and pulse influences from unstated sources). Average velocity of S‐waves in the top 10‐meter thickness changed from 150 m/s up to 400 m/s, and values Vp/Vs – from 2.5 up to 10. In the bottom massif in an interval of depth from 10 m (a level of depth bases of reactor units) up to 20 m from surface, Vs changed from 250 m/s up to 500‐600 m/s at Vp/Vs=2.0–5.5. Increments of seismic intensity (ΔI) in view of position of a level of ground water concerning hypothetical soils II categories on a surface of a site were estimated by size +1 point, at depth on 10 m from surface (from 10 m up to 20 m) – about +0,5 points. Decrements (Dp,s) of seismic waves absorption in the top part of a section were equal in general: for top water saturated soils of 10 m thickness with Vs up to 250 m/s Dp≈1,0, Ds=0,60, for the soils 10 m lying below with Vs>250 m/s Dp=0,50, Ds=0,60. By results of registration of soil fluctuations from various sources (basically man caused) in three points of a

555444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

site with a little bit various engineering‐geological conditions the peak level differed in various items no more than on 20–30%. The prevailing periods of fluctuations, which peak level in 3–4 times were exceeded with microseism, made in the maximal phases 0.15–0.34 with, that as a whole correspond to the resonant periods of fluctuation of thickness of the soils on the site. The constant background of handicapes was characterized by the periods 0.10–0.15 sec. By results of tool supervision by different methods seismic soils conditions of a site can be considered as quasi homogeneous. At calculation of amplification Ratio of thickness of soils on site INPP and designee acceleration of a settlement section (seismological model) up to depth of 700 m were used, as elastic half space the roof of the crystal base was accepted. The basic quantitative characteristics of seismic action from local earthquake according to special calculations under computer programs have made at maximum settlement earthquake for a surface of a site amax=100 sm/sec2, Tamax=0.15‐0.40 sec, effective duration of fluctuations τ≈3–4 sec. In a case of depth the bases on 10 m from a surface amax=60–75 sm/sec2. Seismic conditions on a site for UDBE and DBE from local earthquakes both on a surface, and in a case of bases depth were characterized by a corresponding set settlement and synthesized of accelerograms. For the account of the removed earthquakes such as Carpathian and Skagerrat it was offered to use record accelerograms Carpathian earthquake of 1977, registered 1400 km far from an epicenter, the amplitude scaled on a level up to 50 sm/sec2. On a map of seismic microzonation into districts of site INPP of scale 1:1000 according to complex engineering‐geological and tool researches, and also special calculation the most part of territory (about 70%) at UDBE has been related to seismicity of 7 points. Also the map on a mark of 139 m has been made (a level of depth of reactor units), thus inn calculation of 10 m, lying below specified mark were accepted litological structure and soils properties. In this variant the most part of territory at UDBE was characterized by seismicity of 6.5 points. The approached estimation of an opportunity of deformation of soils has been executed at 7‐units impacts. It was considered that lake‐bog, lake‐glacial sandy loams, loams and clay at dynamic loadings show thixotropic properties, losing structural bonds. According to the approached estimation at 7‐units reactions on a surface of soils cracks on depth up to 5 m can be formed [3]. For 20 years past almost after the researches undertaken on site INPP according to seismic hazard local engineering conditions and seismic properties of soils as a whole have hardly undergone essential changes. However in the basis of reactor blocks under influence of loading and eventually of soils properties could change to the best. Experience of studying constructions, including Rostov atomic power plant, on NPP in Bushehr (Iran) and on other nuclear objects, testify to it [1]. As to an estimation of initial seismicity and initial seismic reactions in view of advanced last year’s probablisting approaches? Including an estimation seismic intensity at repeatability of earthquakes of 1 times in 10000 that is used at UDBE changed in a number of aspects and the requirement of new national normative documents and CODE of IAEA provision concerning a safety of the NPP, including from possible seismic reactions. In view of these changes results of researches of 1987–1988 on site INPP, probably are subject to updating and addition.

555555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

REFERENCES:

Alyoshin A.S., Mindel I.G., Trifonov B.A. Control of seismic properties of soils under the bases reactor units of the nuclear power plants. Geodynamics and technological impact. Materials of the All‐Russia meeting. On September, 12‐15. Yaroslavl. P.13–18. SNiP. Construction in seismic areas. Gosstroy Russia. Moscow, 2002. 45 p. Estimation of influence of soil conditions on seismic danger. A methodical menagement on seismic microzonation. Under edition Pavlov O.V. Moscow, «Nauka», 1988. 223 p.

555666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

AAAPPPPPPRRROOOAAACCCHHH TTTOOO PPPAAARRRAAAMMMEEETTTRRRIIIZZZAAATTTIIIOOONNN OOOFFF TTTEEECCCTTTOOONNNIIICCC EEEAAARRRTTTHHHQQQUUUAAAKKKEEESSS WWWIIITTTHHHIIINNN TTTHHHEEE KKKAAALLLIIINNNIIINNNGGGRRRAAADDD DDDIIISSSTTTRRRIIICCCTTT,,, RRRUUUSSSSSSIIIAAA,,, BBBYYY MMMAAACCCRRROOOSSSEEEIIISSSMMMIIICCC DDDAAATTTAAA

Nikonov A. A. Institute of Physics of the Earth RAS, Bol. Gruzinskaya, 10, Moscow, 123995, Russia, [email protected]

After unexpected Kaliningrad earthquakes on September 21, 2004, scientific activity in the region was developing, among other, into two lines particularly. The first one is assessment of macroseismic data in order to derive main parameters of detected instrumentally three main shocks. And the second purpose consists in searching and treatment according to modern standards initial data on seismic events of the past within the region. The first line has been provided by collective efforts of Russian experts as to concern with three main shocks at 11.05, 14.32, 14.36 GMT September 21. The results of the work basing on macroseismic data have been published [Nikonov, 2006; Nikonov et al., 2006; Aleshin A.S. et al., this issue]. Now the author accomplished additionally parameter estimations for the next four shocks (later aftershocks), which have been registered as sensible by local residents but recorded instrumentally. Due to rather limited relevant information our estimations of time, epicenter location, maximum observed intensity, and derived magnitude are of lesser accuracy in respect to those for the first three shocks. Nevertheless they seems to be useful. Hypocenter depths can be estimated by elements of macroseismic fields (I0, Δi ) for each considered shock. Non‐accuracy is marked by brackets. To derive magnitude values we used the equation I0=1.36MS–2.7lg(h)+3.36 [Assinovskaya, Nikonov, 1988] as more suitable for the region. It should be noted that all calculated meanings of the later aftershock parameters are in good accordance with parameter of the preceded shocks in respect of locality especially, being smaller in intensity and magnitude (Table 1). Table 1. Registered macroseismically earthquakes on 2004 in the Kaliningrad district

№ Date Time Coordinates h, Ms I0 h. m. φ°N; λ°E кm 1 Sent. 21 11 05 54.91; 19.88 15 4.3 VI ±0.08 ±0.02 ±4 ±0.5 2 Sent. 21 13 32 54.97; 20.13 14 4.6 VI–VII ±0.03 ±0.12 ±4 ±0.5 3 Sent. 21 13 36 54.88; 19.91 5 2.2 IV–V ±0.05 ±0.02 ±4 ±0.5 4 Sent. 21 19 30 54.90; 20.05 7 2.3 (IV) ±30 m. ±0.05 (5‐16) ±0.5 ±0.5 5 Sent. 21 21 30 55.0; 20.00 8 2.5 (IV) ±5 m. ±0.10 (5‐12) ±0.5 ±0.5 6 Sent. 22 00 (54.55; 20.10) 5 2.3 (IV–V) ±2 h. ±0.05 (3‐8) ±0.5 ±0.5 7 Sent. 22 (02 55.0; 20.00 (5) 1.8 (III–IV) ±2 h.) ±0.2 (3‐10) ±0.5 ±0.5

So we have now a full list of macroseismically parameterizied felt shocks of the event of 2004 in the Kaliningrad district in order to withdraw relevant conclusions furthermore.

555777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Following the second line of macroseismic research to continue his previous approach [Nikonov, 2005] the author has implemented gathering and thorough analysis of historical descriptive sources on seismic events in the Eastern . The basic sources appeared to be two ones namely chronics of Peter from Duisburg, the author of XV cent. [Peter, 1997], and of later author D. Lucas by name [David, 1813]. Descriptive catalogues of W. Laska [1902] and A. Sieberg [Sieberg, 1940] notes by a couple lines only a seismic event in 1303 mixed with that in Carpathians. Parametric catalogue [Shebalin, Leydecker, 1996] has repeated and confirmed this misunderstanding. Attentive analysis of original texts, being imposed on the geographical and historical realities of that time, allow us to propose position of epicenter areas and some other features of the events. This time we are in position to point out not one but four events (shocks) occurred in 1302 (1303) three of which seemed to be a swarm and fourth one is a separated in place and time event. And one more event is recognized in 1328, i.e. 25 years later in another part of the Eastern Prussia on the Lithuanian border. It was impossible to do magnitude assessment via felt areas because scarce of macroseismic data and particularly great square of aquatorium around. That’s why we have used procedure of magnitude assessment via hypocenter depth. If the latter could be evaluated from derived elements of macroseismic field we realized it by nomogramms from [New catalogue, 1982] using attenuation coefficient ν=3. In other cases conventional depth values 2, 5, 10, 15 km have been used. To calculate magnitude values М (Мmacro or Ms) with better confidence we used several assumptions firstly and some equations of macroseismic field of regional sense. They are following

I0=1.36MS–2.7lg(h)+3.36 (I) [Assinovskaya, Nikonov, 1988]

Ms=0.65I0+1.9log(h)–1.62 (II) [Kaizer et al., 2002]

ML=0.88I0+0.64log (h) –1.52 (III) [Grünthal, Wahlstrom, 2003]

In general we prefer to use the equation (I) as being checked out recently for the three main Kaliningrad shocks on September 2004 [Nikonov et al., 2007, this issue]. Deflections in given results have been taken into account finally by procedure of disperse estimations for each event. Of course, we have quite insufficient data to evaluate surely so old event parameters and our estimations have to be considered as having not high reliability (so they are presented in brackets) (Table 2). Table 2. Earthquakes in the Kaliningrad district (Eastern Prussia) in the beginning of XIV cent.

№ Date Time Coordinates H, Ms I0 Notes h. m. φ°N; λ°E кm 1 1302 (1303) 10 30 55.0; 20.0 (15) (4.5) VI–VII Strength of maximal among Aug. 15 ±2 h. ±0.5 ±1.0 10‐20 ±0.5 ±0.5 three shocks is evaluated 2 1302 (1303) 10 30 55.0; 20.0 (15) (4.3) (V–VI) Conditional evaluation in Aug. 15 ±2 h. ±0.5 ±1.0 10‐20 ±0.5 ±0.5 respect of strongest shock 3 1302 (1303) 10 30 55.0; 20.0 (15) (3.8) (V–VI) Aug. 15 ±2 h. ±0.5 ±1.0 10‐20 ±0.5 ±1 4 1302 14 00 55.3; 21.0 (10) (5.0) (VII) Conditional evaluation by (1303) ±6 h. ±0.3 ±0.5 5‐20 ±0.5 ±0.5 water waves high Dec.1 ±30 days 5 1328 55.1; 23.5 (10) (5.0) (VII) ±0.3 ±0.5 6‐18 ±0.5 ±0.5

555888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Comparing separated in time 700 years about two bursts of regional seismic activity in the beginnings of XIV and XXI centuries we can bring to light some features of remarkable similarity. So in this step of our knowledge we are in position to assess general seismic situation in the region in better way in order to speculate more about long‐term seismic danger in the Kaliningrad district and in its vicinity.

REFERENCES:

Aleshin A.S., Aptikaev F.F., Nikonov A.A., Pogrebchenko V.V. Discrepancy in location of macroseismically and instrumentally derived epicenters of Kaliningrad, September 21, 2004, earthquake – factual data and possible explanations / This issue. 2007. Assinovskaya B.A., Nikonov A.A. Felt earthquakes of the XXth century in the Eastern Baltic shield / XXVI General Assembly of the European Seismological Commission: Abstracts. Tel Aviv, Israel, 23‐28 Aug., 1998. P. 10. David M. Lucas. Preussische Chronik. Fünfter Band. Königsberg, 1813. 246 s.

Grünthal G., Wahlstrom R. An Mw based earthquake catalogue for central, northern and using a hierarchy of magnitude conversions // Journal of Seismology. 2003. V. 7. P. 507‐531. Kaizer D., Gutdeutsch R., Jentsch G. Estimating the Magnitude Ms of Historical Earthquakes from Macroseismic Observation / Geophysical Research Abstracts. V. 4. 2002. 27th General Assembly. NH4. EGS02‐A.–02495. Laska W. Die Erdbeben Polens. Mitt. Erdbeben‐Comission d. K. Akademie d. Wissenschaften. Neue Folge. № 8. Wien. 1902. 36 s. New catalogue of strong earthquakes in the USSR from ancient times through 1977. Eds.‐in‐Chief N.V. Kondorskaya and N.V. Shebalin. Boulder. 1982. Nikonov A.A. Were there worthy of note earthquakes and tsunamis within the south‐eastern Baltic area? / In: Joeleht (ed.) Kaliningrad Earthquake September 21, 2004: Workshop Materials. Tartu, 2005. P. 23‐25. Nikonov A.A. Source Mechanism of the Kaliningrad Earthquake on September 21, 2004 // Doklady Earth Sciences. 2006. V. 407, N 2. P. 317‐320. Nikonov A.A., Aptikaev F.F., Aleshin A.S., Assinovskaya B.А., Pogrebchenko V.V. Kaliningrad earthquake of September 21, 2004, as a model one for the East European platform / 21th Century Geophysics: 2006. Papers of the 8th V.V. Fedynski Gephysical Conference. М. 2006. P. 282‐289. (In Russian). Nikonov A.A., Pacesa A., Aptikaev F.F., Aleshin A.S., Pogrebchenko V.V. Kaliningrad, September 21, 2004, earthquake in the Eastern Baltic area basic macroseismic maps for three main shocks / This issue. 2007. Peter iz Duisburga. Khronika zemli Prusskoi (Peter from Duisburg. Chronic of the Prussia Land). М.: Ladomir, 1997. 384 p. (In Russian). Shebalin N.V., Leydecker. G. Earthquake catalogue for Central and Southeastern Europe 342 BC‐1990 AD. Moscow‐Hannover. 1996. 195 p. Sieberg A. Beitrage zum Erdbebenkatalog Deutschlands und angrenzender Gebiete fur Jahre 58 bis 1799 / Mitt. Deutsch. Reichs‐Erdbebendienst. 2. 1940. 112 s.

555999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

KKKAAALLLIIINNNIIINNNGGGRRRAAADDD,,, SSSEEEPPPTTTEEEMMMBBBEEERRR 222111,,, 222000000444,,, EEAAARRRTTTHHHQQQUUUAAAKKKEEE IIINNN TTTHHHEEE EEEAAASSSTTTEEERRRNNN BBBAAALLLTTTIIICCC AAARRREEEAAA ––– BBBAAASSSIIICCC MMMAAACCCRRROOOSSSEEEIIISSSMMMIIICCC MMMAAAPPPSSS FFFOOORRR TTTHHHRRREEEEEE MMMAAAIIINNN SSSHHHOOOCCCKKKSSS

Nikonov A.A.1, Pačėsa A.2, Aptikaev F.F.1, Nikulin V.G.3, Puura V.4 and Aronov A.G.5 1Institute of Physics of the Earth, Russian Academy of Sciences, Moscow, Russia, [email protected], [email protected] 2Lithuanian Geological Survey, Vilnius, Lithuania, [email protected] 3Latvia Environment, Geology and Meteorology Agency, Riga, Latvia, [email protected] 4University of Tartu, Estonia, [email protected] 5Center of Geophysical Monitoring, Minsk, Belarus, [email protected]

The earthquakes on September 21, 2004 aroused in the territory of the Russian Kaliningrad enclave and were felt as far as Norway and Belarus (Wiejacz et al., 2006). The first event occurred at 11:05 UTC, the second at 13:32 UTC, and a small aftershock followed the second event four minutes later. The magnitudes of these events were mb=4.4, mb=5.0, mb=3.0, and epicenter intensities Io=6, Io=6.5, Io=4, respectively. Intensity maps for the near‐epicenter area for the two largest earthquakes were published (see Nikonov, 2006; Nikonov et al., 2006). The differences in the patterns of the maximum intensities of all three earthquakes indicate slightly different source locations offshore very close to the coasts of the Sambia Peninsula. The collection of macroseismic data was carried out country‐wise within the perceptibility area, but intensity assessment was discussed and coordinated generally in a workshop in Tartu, Estonia in May 2005 (Jõeleht, 2005). Personal contact to sources of information has supplemented questionnaires and photographs in many places. The European Macroseismic Scale 1998 (EMS) was chosen for evaluation of the intensity observations. This work was continued between the main representatives of Estonia, Latvia, Lithuania and Belarus under Russian supervision with special attention to cross‐boundary correlation and adjustment of initial data. As a result, maps for the three largest events have been compiled and isoseismals 2‐6 contoured for the Eastern Baltic area in 2005. These maps are publishing here for the first time. The macroseismic field equation I = a MS – b lg (h2 + RI2)0.5 + c allow to estimate all the coefficients of equation and parameter values. Here are: MS – surface wave magnitude, h – foci depth, RI – epicenter distance to isoseismal I, a, b, c coefficients. In the case considered it’s difficult to determine earthquake parameters because of rather complicate shapes of isoseismals. To reduce the errors the nearest and the distant isoseismals they weren’t taken into account. The measurements were carried out by using isoseismals I = 4.5 and I = 2.5 (ΔI = 2). The most unstable coefficient b is described by the intensity attenuation: ΔI (ΔI=2) = b lg ((h2 + R4.52)0.5/(h2 + R2.52)0.5)). The obtained values correspond to directions N (to Ventspils), NE (to Riga), E (to Vitebsk) SE (to Masyr) and S (to Warszawa). The foci depth as first iteration is taken equal 17 km (Malovichko et al., 2006), that is much less than epicenter distances used. The estimations of coefficient b, obtained for the three shocks are in good agreement (table).

666000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Table. The coefficient b for different directions

Directions Earthquake N NE E SE S 1 2.25 2.1 3.25 2.25 3.9 2 2.3 2.0 3.1 2.0 3.9 3 – 2.0 – 2.1 – Mean 2.3 2.0 3.2 2.1 3.9

The average value for three shocks for all the used directions is b = 2.70 ± 0.76, i.e. the same as for the Fennoscandian shield periphery area (Assinovskaya, Nikonov, 1998). This value significantly differ from mean world‐wide value (b = 3.5). The most complicated macroseismic field is observed by largest aftershock. The intensity difference can be up to 3 intensity units at the neighboring points, my be because of very high frequency of oscillation or due to different characteristics of soil. Therefore estimations of attenuation coefficient for the aftershock are possible for two directions only. The largest attenuation is observed in E – and S – directions, and the weak one is in NE – and SE – directions. Other coefficients of equation are much more stable. Therefore, the equation: I = 1.36 MS – 2.7 lg (h2 + R2)0.5 + 3.36 was used to evaluate magnitudes (Assinovskaya, Nikonov, 1998). For the first shock mean distance using five directions is R2.5 = 310 km and one can obtain MS = 4.3. For the main shock R2.5 = 413 km and MS = 4.6.

Figures. Macroseismic maps of all three Kaliningrad shocks. The upper one – 11:05 UTC shock, the one on the right – 13:32 UTC and the one on the next page – 13:36 UTC. Number of actual intensity points for Kaliningrad enclave and Lithuania was significantly reduced in order to avoid overlapping of figures.

666111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

In order to evaluate focal depths the same equation was modified to fit epicenteral area: Iо = 1.36 MS – 2.7 lg h + 3.36. The epicenters are located about 5 km from coastline for both of shocks, the intensity may be practically the same as largest observed on coast area, and h = 15 km for the first shock and h = 14 km for the second one. In our opinion, very pronounced polarization of seismic intensity attenuation rather similar for three shocks generally should be explained by earthquake source mechanisms and inhomogeneity of geological structure of the felt area. The question is initially considered in [Nikonov, 2006a, b]. REFERENCES: Assinovskaya B.A., Nikonov A.A. Felt earthquqkes of the XXth century in the Eastern Baltic shield // XXVI General Assembly of the Europ. Semsmol. Commiss.: Abstr. Tel Aviv, Israel, 23‐28 Aug., 1998. P. 10. Joeleht A. (ed.), 2005. The Kaliningrad earthquake September 21, 2004. Workshop Materials. Institute of Geology, University of Tartu. Tartu, 52pp. Malovichko A.A., Mekhrushev D.Yu., Starovoit O.E., Gabsatarova I.P., Chepkunas L. S. On Kaliningrad September 21, 2004 earthquake and establishing seismic monitoring in the Kaliningrad district, Russia / Modern methods of processing and interpretation of seismological data. Materials from International seismological school dedicated to 100‐ anniversary foundation of seismic stations “Pulkovo” and “ Ekaterinburg”. Peterhof, October 2‐6, 2006. PP. 89–97. (In Russian). Nikonov A.A. Source mechanism of the Kaliningrad earthquake on Sptember 21, 2004 // Doklady Earth Sciences. 2006a. Vol. 407, N 2. PP. 317‐320. Nikonov A.A. Kaliningrad, 2004, earthquake on platform area – problems under consideration and those to be resolved // Active geological and geophysical lithospheric processes. Methods, technics and results of studying. Materials of the XII International conference on September 18‐23, 2006. Voronezh. 2006b. Т. 2. P. 21–22. (In Russian) Nikonov A.A., Aptikaev F.F., Aleshin A.S., Assinovskaya B.A., Pogrebchenko V.V. Kaliningrad September 21, 2004 earthquake as model for the East Europe platform // 21st Century Geophysics: 2005. Papers of the 8th. V.V. Fedynski Geophysical Conference Мoscow: Nauchny mir, 2006. PP. 282–289. (In Russian). Wiejacz P., Nikonov A.A., Gregersen S., Aptikaev F.F., Aleshin A.S., Pogrebchenko V.V., Debski V., Assinovskaya B.A., Guterch B., Pacesa A., Mantyniemi P., Nikulin V.G., Puura V., Aronov A.G., Aronova T.I., Grunthal G., Huisby E.S., Sliaupa S. Exceptional earthquakes of September 21, 2004 in the Kaliningrad district, Russia // Modern methods of processing and interpretation of seismological data. Materials from International seismological school dedicated to 100‐anniversary foundation of seismic stations “Pulkovo” and “ Ekaterinburg”. Peterhof, October 2‐6, 2006. P. 103–107 (In Russian).

666222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

RRREEEGGGIIIOOONNNAAALLL FFFEEEAAATTTUUURRREEESSS OOOFFF SSSEEEIIISSSMMMOOOTTTEEECCCTTTOOONNNIICCCSSS AAANNNDDD DDDEEEFFFOOORRRMMMAAATTTIIIOOONNN OOOFFF EEEAAARRRTTTHHH CCCRRRUUUSSSTTT OOOFFF TTTHHHEEE BBBAAALLLTTTIIICCC RRREEEGGGIIIOOONNN

Nikulin V. G. Latvian Environment, Geology and Meteorology Agency, Riga, Latvia, [email protected]

Attitude toward seismicity of countries of south‐east part of the Baltic region (Lithuania, Latvia, Estonia, Kaliningrad area of Russia) substantially was changed after the earthquakes of 1976 on Osmussaar Island in Estonia (Mw = 5.0) and 2004 in the Kaliningrad area of Russia (Mw = 5.3). In connection with the existent and projected objects of energy, including atomic (Ignalina NPP), and by the systems of pipelines in the countries of Baltic and in the Baltic Sea (http://en.wikipedia.org/wiki/Nord_Stream), it is necessary to take into account a credible seismic danger and danger of deformation of those areas of the earthʹs crust, where is assumed to place these objects. Seismotectonic analysis is difficult for territory of the Baltic region, which is located on the north‐west of the East‐European platform. It is linked with the limited seismic statistics, lot of man‐made seismic events, which it is necessary to identify and exclude from the analysis of seismotectonic danger, insignificant number of the seismic stations and limited time of regional instrumental observations. In these terms, the role of geological and geophysical indicators of seismicity is increased for description of seismic sources zones in which earthquakes happened already. The discovered signs of seismicity can be used for recognition of other potential seismic sources zones. One of indicators of seismicity there can be areas with intensive neotectonic (35– 37 million years – Rupelian) motions of the earthʹs crust. The Baltic system of grabens of the submeridional extending is located in the Baltic region, which intersects with the Finnish system of grabens of the sublatitudinal extending (Garetsky et al., 1999). Gdansk depression, East Gotland and Finnish grabens form the depressed system, which occupies central part of the Baltic Sea. The most active parts of neotectonic structures are borders of grabens and depresion zones (fig. 1). The maximal horizontal gradients of neotectonic motions can be perspective characteristics for detection of seismic sources zones.

Fig. 1. Total amplitudes of neotectonic motions of the earthʹs crust and epicenters of earthquakes of the Baltic region (1375–2006) 1 – lines of equal values of total neotectonic amplitudes in meters, 2 – area of neotectonic depressions of the Baltic Sea and coast of Baltic countries, 3 – epicenters of earthquakes (the diameter of circumference corresponds the size of moment magnitude Mw), 4 – areas of the maximal accumulated seismic moment M0.

666333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

The depression zone in central part of the Baltic Sea is characterized lowering of the earthʹs crust with maximal amplitude more than 280 m during the neotectonic stage. Amplitudes of raising (near + 80 m) are belonging to Zemaitia – Kurzeme and Vidzeme highs. More than 290 earthquakes happened within the limits of the indicated region of the Baltic Sea (Δϕ = 53.75°N ‐ 60.45°N and Δλ = 15.35°E ‐ 30.45°E), since 1375 to 2006 year. The catalogue of earthquakes of the Baltic region is made on the basis of regional catalogues (Ahjos and Uski, 1992; Avotinya and other, 1988) and more late updates http://www.seismo.helsinki.fi/bul/index.html. In the Baltic region the areas of display of seismicity are often belonging to the areas of horizontal gradient of total amplitudes of neotectonic motions (gradNHMSA). Near‐by the seismic source of the Kaliningrad earthquakes of 2004 of gradNHMSA achieves 3.7 m/km, and near‐by the seismic source of Osmussaar earthquakes of gradNHMSA are 3.6 m/km. Epicentres of earthquakes in Pyarnu in 1670 (Mw = 3.9), in Liepaja in 1909 (Mw = 3.5), in Ventspils in 1785 (Mw = 3.1), south‐east and north‐east of Gdansk in 1994 and 2004 (Mw = 2.4 and 2.6), to the south of Gotland island in 2002 (Mw = 3.1) also located in the areas of promoted gradNHMSA. It is certificate about the inherited character of deformation of the earthʹs crust along the edges of neotectonic structures.

The seismic moment M0, which well correlated with the sizes of seismic sources of earthquake, was used (Kagan, 2002) for the analysis of seismotectonic activity of the earthʹs crust. M0 characterizes complete energy of deformation (work), realized in the process of earthquake (Kanamori, 1977). Seismic energy ES is only small part (5×10–5) of general energy of deformation and emanates from the hearth of earthquake as seismic waves. M0 is considered as an index of deformation of the earthʹs crust. Comparison of seismic moment (conditional deformations) of the earthʹs crust of regions of platforms, which have general conditions of geological development, allows to do generalizations and discover regional features. The indexes of deformation of the earthʹs crust of the Baltic region and his part (contour on fig.1) were compared to the similar indexes of district of bay of Hudson on the North‐America platform, which on the sizes almost in 3 times excels the area of the Baltic sea. Postglacial raising in the district of bay of Hudson in last 6 thousands of years were attained by more than 120 m (Walcott, 1975). For the Baltic region and south Scandinavia (B&SS), part of the Baltic region (B) and district of bay of Hudson (H) is got exponential dependence of the accumulated seismic moment 10 0.007T ( ∑ M 0 ) from time of accumulation (T): ∑ M 0B&SS = 8×10 × e , 11 0.006T −11 0.156T ∑M 0B = 3×10 × e , ∑ M 0H = 1×10 × e .

The rate of the accumulated deformation Vε can to characterise the dynamics of

seismotectonic process. It is expressed through the accumulated seismic moment ∑ M 0 on the area unit (1 km2) of the earthʹs crust in time unit (1 year):

Vε = (∑∑M 0E − ∑ M 0I )/( M 0I × ΔT × S) (1)

where, ∑ M 0E and ∑ M 0I are sums of seismic moment in the end and at the beginning of certain time domain, DT is a time of accumulation of deformation (sum of seismic moments) domain, S is an area, km2.

666444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

The rate of the accumulated deformation allows to consider seismotectonic activity of different regions regardless of sizes of area of researches. For an analysis was chosen time interval 30 years from 1974 to 2004. The maximal moment magnitude Mw equal 6.3 for earthquake in 1969 in the district of bay of Hudson. That defined the sharp change of the graph of accumulated seismic moment M0.

The rate of deformation Vε for the Baltic region and for the district of bay of Hudson appeared practically identical (9.4×10–9×(1/(year×km2)), and for part of the Baltic region (contour on a fig. 1) it almost in 2 times higher 2.1×10–8× (1/( year×km2).

Conclusions:

1. Kaliningrad, Stockholm, Osmussaar and Bauska areas of earth crust on which the sum of seismic moments achieves maximal sizes 2.7×1016 – 2.6×1017 Nm are located in the Baltic region. Three first areas of the earthʹs crust are located in the Baltic Sea. 2. Rate of the accumulated deformation (by seismic moments) of the earthʹs crust for the region of the Baltic Sea and bay of Hudson are practically comparable and are equal 9.4×10–9×(1/(year×km2)). Rate of deformation of the earthʹs crust 2.1×10–8× (1/(year×km2) for system of Baltic grabens, Gdansk depression and coast of Baltic countries (contour on fig.1) is approximately in 2 times more than in all Baltic region and in the district of bay of Hudson. 3. Kaliningrad (Mw = 5.3) and Osmussar (Mw = 5.0) of earthquakes sources belong to the areas of the earthʹs crust with the maximal values of horizontal gradients of neotectonic motions (3.6–3.7 m/km). Neotectonic structures edges are seismically more active, than their middle part. Earthquakes sources belonging to the areas of the earthʹs crust with the high values of horizontal gradient of neotectonic motions. It shows that in the Baltic region on the edges of neotectonic structures there is modern deformation of the earthʹs crust.

REFERENCES:

Ahjos T., Uski M. Earthquakes in in 1375–1989. // Tectonophysics, 1992, 207, 1–23. Авотиня И.Я., Боборыкин А.М., Емельянов А.П., Сильдвээ Х.Х. каталог исторических землетрясений Белоруссии и Прибалтики // Сейсмологический бюллетень сейсмических станций «Минск» (Плещеницы) и «Нарочь» за 1984 год. ОНТИИ, Минск, 1988, 126–137. Garetsky R., Levkov E., Schwab G., Karabanov A., Aizberg R., Garbar D., Kockel F., Ludwig A.O., Lukke‐Anderson H., Ostaficzuk S., Palienko V., Sim L., Sliaupa A., Sokolovski J., Stackebrandt W. Main neogeodynamic features of the Baltic Sea depression and adjacent areas // Technika poszukiwan geologicznych. Geosynoptyka i geotermia. Proceeding of the 7‐ th International meeting of the working group Neogeodynamica Baltika. 2‐6 June, 1997. Special issue. In: Ostaficzuk S. et al., (eds), Krakov, 1999, 1, 17–27. Kagan Y.Y. Aftershock zone scaling // Bulletin Seismological society of America, 2002, v.92, 641 – 655. Kanamori H. The energy release in great earthquakes // Journal of Geophysics research, 1977, v.82, Nr. 20, 2981–2987. Walcott R.I. Vharacteristics of recent uplift in // Problems of Recent Crustal Movements. Valgus, Tallinn, 1975, pp. 146–155.

666555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMOOOLLLOOOGGGIIICCCAAALLL OOOBBBSSSEEERRRVVVAAATTTIIIOOONNNSSS IIINNN LLLIIITTTHHUUUAAANNNIIIAAA

Pačėsa A. Lithuanian Geological Survey, Vilnius, Lithuania, [email protected]

Territory of Lithuania and whole region of Eastern Baltic feature a low seismic activity. Earthʹs crust of early Precambrian consolidation and significant distances to active tectonic zones causes situation of this kind. Nevertheless, according to historical and instrumental data a few dozens of local earthquakes with intensities reaching VII points (MSK scale) took place in the Baltic countries and adjacent Belarus since 1616 to our days (figure 1). Two Kaliningrad earthquakes with magnitudes 4.5 and 5.0 stroke Baltic region in 2004 which indicated seismogenic potential of this region. Besides manifestation of some local seismic activity in Eastern Baltic, large regional earthquakes generate earth trembling up to intensities IV or V (MSK scale) in this area. For instance, inhabitants of Lithuania have felt trembling from Oslo 1905 earthquake and from earthquakes of Vrancea area in Romania in years 1940, 1977, 1986 and 1990. The first instrumental seismological observations in Lithuania started in 1970 as Vilnius seismic station was founded. Three analog long period (T=25 s) and three short period (T=1.5 s) seismometers were installed in the territory of Institute of Physics at outskirts of Vilnius. Seismological records were sent to Obninsk (Russia) seismological centre and processed there until year 1992. Later on, maintenance of station and routine data processing was undertaken by stuff of Institute of Physics. 450 distant and regional seismic events were reported in the seismic bulletin of Vilnius seismic station since 1991 to 1995. Phase readings, amplitudes and their periods were defined during routine processing. Distances from the epicentres to Vilnius station were calculated and reported in the bulletin. No local events were registered in Vilnius seismic station. Possibly, high‐level background noise at this site reduced resolution of short period seismometers inhibiting detection of small local events. Operation of Vilnius seismic station was suspended in the beginning of 1999. The fist comprehensive study of seismic activity of Lithuania was carried out in 1988 as a part of re‐examination of safety of Ignalina Nuclear Power Plant (INPP). The top twenty‐two experts of the concluded that seismic hazard was not assessed properly when INPP has been designed despite local and international regulations. In order to improve the situation the experts proposed to install seismic network and monitor local seismicity. Seismic Alarm System (SAS) and complementary Seismic Monitoring System (SMS) were installed in the INPP in 1999. The SAS was designed to issue alarm when damaging seismic wave approaches the INPP, while the SMS was designed to collect data of local seismicity and dynamic behaviour of structures of the INPP. According to the design, the SMS had to consist of six vertical short period seismometers installed in boreholes at the depth of 30m and located at a distance of ~30 km to the INPP. The signals of the seismometers had to be transformed to digital form and telemetered continuously to the control centre at the INPP. The control centre recorded data of seismometers if an event had been detected. Unfortunately, only four seismic stations were built on territory of Lithuania. Due to some political misunderstandings, seismic stations on the territories of Latvia and Belarus haven’t been built. At the same time Geological Survey of Lithuania took responsibility to process, analyse and store seismological data of the SMS and project of seismological monitoring was initiated there. Processed data of the SMS, analyses of seismological bulletins of NORSAR and

666666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Seismological institute of Helsinki University (HU) related to Lithuania’s territory and some other results have been presented in the annual seismological bulletins and reports of LGS (e.g. www.lgt.lt/seismo). Copies of the seismological bulletins were sent to ISC (International Seismological Centre) regularly ant to EMSC (European Meditiranian Seismological Centre) lately. Results of the first stage of project of seismological monitoring (year 1999–2005) supported an idea of low seismic activity of the territory of Lithuania and adjacent areas. Just a few local tectonic events have been registered by the seismic system of INPP during this period (figure 2). None of them was located in Lithuania. Analysis of data of seismic bulletins of NORSAR and Seismological Institute of HU has shown the similar results (figure 3). Seismic bulletins of NORSAR and HU reported approximately 50 explosion events located in the territory of Lithuania and nearby areas each year. Vast majority of them could be associated with quarry blasts and some military exercises. Meanwhile, just a very few events of this kind have been registered by the SMS of INPP and a few probable explanations of this situation were found recently. Nevertheless, earthquakes with magnitudes above 5.0 can disturb long periods of quietness in this area, as Kaliningad seismic events have demonstrated. Earth trembling of Kaliningrad events was felt on the large part of territory of Lithuania and significant amount of efforts was dedicated collecting and processing of macroseismic data during a few years after events occurred.

Figure 1. Map of earthquakes of Eastern Baltic since 1616 to present. Shaded circles correspond to historical seismic events, hexagons – instrumenticaly recorded events, triangles – operating seismic stations (a few newly installed seismic stations are not depicted).

666777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Figure 2. (Next page, figure at the top) Seismic events registered by Ignalina SMS since December 1999 till June 2005. Circles correspond to tectonic seismic events, crosses – explosion events, open triangles – quarries where explosions could be executed, filled triangles – seismic stations of Ignalina NPP, dashed line – line of deep sounding experiment CELEBRATION, squares – explosion points of CELEBRATION experiment, area confined by dashed line – presumptive location of local seismic event of 2001.

Figure 3. (Next page, figure at the bottom) Seismic events reported in Helsinki University bulletin covering the time period since December 1999 till June 2005. Circles correspond to seismic events, crosses – explosion events, grey triangles – quarries where explosions could be executed, filled triangles – seismic stations of Ignalina NPP, dashed line – line of deep sounding experiment CELEBRATION, squares – explosion points of CELEBRATION experiment.

666888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

AAAPPPPPPLLLIIICCCAAATTTIIIOOONNN OOOFFF TTTHHHEEE PPPRRROOOBBBAAABBBIIILLLIIISSSTTTIIICCC AAPPPPPPRRROOOAAACCCHHH IIINNN AAASSSSSSEEESSSSSSMMMEEENNNTTT OOOFFF TTTHHHEEE SSSEEEIIISSSMMMIIICCC HHHAAAZZZAAARRRDDD OOOFFF TTTHHHEEE BBBAAALLLTTTIIICCC RRREEEGGGIIIOOONNN

Pačėsa A.1, Šliaupa S.2 1Lithuanian Geological Survey; [email protected] 2Institute of Geology and Geography; Vilnius University

There are different approaches in assessment of the seismic potential of the particular tectonic regions and structures. Deterministic (DSHA) and probabilistic seismic hazard analyses (PSHA) are two different methods commonly used in seismic risk studies. DSHA is based on geological specific of the site, whereas PSHA is focused on earthquake statistics and numerical calculations. DSHA is more reliable than PSHA because it considers observed geological features and is more transparent. DSHA evaluates earthquake hazards reliably from the geology regardless of time and has no need for time‐based probability. The seismological information that provides a base for the probabilistic analysis, in the Baltic region is rather poor because of (i) low seismic activity, and (ii) insufficient seismological monitoring system. Therefore, the previous studies were based on deterministic‐oriented approaches. Still they also lack the consistency due to geological specifics (low amplitudes of tectonic structures, complicated Quaternary cover, and other factors that limit the resolution the geophysical surveys), scarce geological and geophysical information. As an alternative to the previous praxis, the probabilistic approach was applied in the presented study to assess the seismic potential of the Baltic region. It was urged, first of all, by the need of adaptation of the proved methodologies for derivation of the seismic load for SL‐2 (highest level) seismic risk assessment of the Ignalina NPP (Sliaupa et al., this volume). Seven seismic sources (zones) were distinguished in the Baltic sedimentary basin based on the seismological and geological information, i.e. Leba, Eat Baltic Sea, Kaliningrad, West Estonian, East Baltic North, Latvian, and East Baltic South that crosses the Ignalina NPP. The seismic attenuation equation by Ambraseys et al. (2005) was applied. It is important that it accounts for the soil type and the faulting mechanism. CRISIS 99 software was employed to perform the probabilistic analysis of the seismological information. The seismological catalogue was prepared based on the historical data and scarce instrumental seismic observations. NORDIC format was used for the catalogue, while programs from SEISAN 8.1 software suite were employed for data transformations. The intensity of the historical earthquakes was recalculated to the magnitudes using the equation derived from the West Carpathian region (M=0.55×I0+0.95). The estimated b parameter for the Baltic region is 0.63 (for comparison, it is 0.59 for Finland). The Peak Ground Acceleration (PGA) was calculated for the probability of exceedens 10‐4 that is typical value for SL2 seismic risk assessment. The estimated PGA of the East Baltic zone where Ignalina NPP is located was assessed as high as 0.13 g. The potential of the most representative (in terms of the seismological information) Latvia source was assessed PGA=0.166 g. Similar values were estimated for the other seismic sources in the Baltic basin. For example, in Osmussaare area PGA exceeds 0.2 g, in Kaliningrad area PGA is up to 0.17 g. Most of territory of Lithuania, in this model, is characterized by very low seismic hazard which is due to lack of seismic events recorded in this area.

666999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

It is worth mentioning, that the largest uncertainties came from ambiguous definition of the size and form of seismic zones. As certain amount of energy of seismological activity was confined within smaller area, the greater PGA values were obtained for the certain seismic zone. Therefore, special care should be addressed to delineation of the seismic sources in this area. It can be concluded that the application of the probabilistic approach can significantly contribute to the assessment of the seismic hazard of the Baltic territory. However, it has some important shortcomings that should be realized when interpreting the model results.

Seismic hazard map of the Baltic region, PGA, probability of exceedens 10–4

777000 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

IIINNNSSSTTTRRRUUUMMMEEENNNTTTAAATTTIIIOOONNN OOOFFF TTTHHHEEE SSSEEEIIISSSMMMIIICCC AAALLLAARRRMMM AAANNNDDD MMMOOONNNIIITTTOOORRRIIINNNGGG SSSYYYSSSTTTEEEMMMSSS OOOFFF IIIGGGNNNAAALLLIIINNNAAA NNNUUUCCCLLLEEEAAARRR PPPOOOWWWEEERRR PPPLLLAAANNNTTT

Razinkov O.1, Epp M.1, Kündig C.1, Davidiuk O.2, Narbuntas J.2 1GeoSIG Ltd., Europastrasse 11, 8152 Glattbrugg, Switzerland; [email protected] 2State Enterprise Ignalina Nuclear Power Plant, 31500 Visaginas, Lithuania

The Seismic Alarm (SAS) and Seismic Monitoring Systems (SMS) are in operation at Ignalina Nuclear Power Plant (INPP) since July 1999. Seismic Alarm System is intended to provide the seismic protection of INPP by issuing alarm signals in the main control room, if a seismic wave reaches one or several outside seismic stations, which are located in several dozens of kilometres from main INPP building. These alarm signals may be used to initiate the shut‐down of the reactor by the Fast Acting Scram System (FASS). Considering the time required by the FASS, the shut down of the reactor is feasible before the arrival of the potentially damaging waves of a strong earthquake at INPP. By this way, the risk of radioactivity release to the environment at a seismic event could be strongly reduced. The SMS provides records of small earthquakes or teleseismic events. This promotes the knowledge regarding the seismic activity of Lithuania and the Baltic region. The data recorded by four outside stations of the SMS are delivered to the Lithuanian Geological Survey, analysed and stored there is the database of seismic events. Four outside seismic stations are installed at a distance of approximately 30 km from the power plant, forming a seismic ‘fence’. At each outside station of the SAS, three triaxial accelerometers are installed. One accelerometer is installed in a cabin at the actual location of the outside station and two are in cubicles at a distance of approximately 500 m from the cabin. This provides redundancy to the system: in case of malfunctioning of one sensor, two sensors are still working, which allows the application of a 2‐out‐of‐3 logic selection. Moreover, this arrangement is less susceptible to environmental noise caused by e.g. road traffic and to any chance impact factors because two distant sensors are unlikely affected by such factors simultaneously. At the outside stations, the cabin and the cubicles are interconnected by cables. The cable trenches house the signal and the power cables. The digitizers are installed inside of accelerometer housing and therefore no analogue signals are transmitted in the cable trenches. Analogue signals would be influenced by eddy currents induced by the power supply running in the same cable trench. Seismic equipment installed in the cabin has a function of the seismic switch: it determines whether the seismic signal received from accelerometers exceed a certain threshold and activates a discrete seismic status signal in this case. Every outside station has also the uniaxial seismometer located in a borehole with a depth of 30 m. The digitized signals of these seismometers, together with the seismic status and diagnostic information, are transmitted continuously to INPP by the radio telemetry. The signals received in INPP are evaluated according to a 2‐out‐of‐3‐logic and the alarm is issued in case of an earthquake. Both SAS and SMS subsystems have been operating for more than six years in terms of their basic functionality. Nevertheless, during this period of time several weak points of the original design and the actual functionality were realised and the areas of improvements were

777111 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

identified. Therefore in the year 2006 GeoSIG was asked to perform the modernisation of the system within the frames of a new contract. This work is currently going on. The solutions being implemented during modernisation are based on the best GeoSIG experience collected in such large projects as monitoring of the Øresund Bridge1, Early Warning and Rapid Response seismic system in Turkey2, seismic instrumentation of the tunnels and Swiss nuclear plants3, etc. The main improvements imply the reviewed alarm interface and a modern alarm display board in the control room providing detailed information to the operators about the system status, fully upgraded data collection and processing system in the centre, and the new computerised data acquisition systems at all four outside stations. In order to improve the functionality and reliability of SAS and SMS, GeoSIG is implementing a conceptually new design of the system based on the intellectual acquisition and processing of the seismic data. Buffering and pre‐processing data at outside stations combined with the advanced analysis of the seismic information arriving to the centre allow full compliance with the main requirements on the system modernization. The reliability and performance of the computer technique was increased significantly during last years, which made it possible to use computerised systems in such critical processes that require high level of the reliability. Since the modernization of the seismic system allows bigger amounts of data to be delivered reliably to the centre, the triaxial short period seismometers VE‐53 with the sensitivity of 1000 V/m/s are installed additionally in all outside stations. Unfortunately, the SMS part of the system was assigned significantly less importance than SAS in the original design and therefore three of four outside stations were placed in the sites with relatively high level of ambient noise that makes precise seismic measurements rather problematic. Therefore the installation of more sensitive seismometers was not considered. Nevertheless, the improved instrumentation would allow detection and recording of many more seismic events, including those caused by the industrial activity throughout the monitored area. It is supposed that all digitized data provided by triaxial seismometers will not be transferred to INPP in real time. It is buffered instead in the outside stations for several days so the required fragments of records can be downloaded by operator’s request from the centre. Automatic event detection using the STA/LTA criteria is also working at the outside stations. The recorded events are transferred to the centre automatically within several minutes after an earthquake. Operation of all outside stations and the central system installed in main control room of INPP is based on the multi‐purpose software GeoDAS (GeoSIG Data Acquisition System). This software has been used in many projects during last six years and it is proven to be handy, stable and reliable. GeoDAS has been designed to meet all requirements with respect to almost every possible data acquisition application. The program has an open architecture not only for multiple local recorders, but also for networks of remote recorders, supporting different types of communication, including links via Internet. It includes the comprehensive viewer for reviewing and interactive analysis of the digital signals off‐line. Data channels are displayed in a way that operators can see on the screen and plot any set of them in any combination, make scaling, zooming, axis style changing, export to and import from various data formats in an intuitive manner. Besides of the standard functionality, GeoDAS supports also some specific applications and complex data acquisition and telecommunication systems. In particular, several software modules are developed and added to GeoDAS in order to meet INPP requirements on modernisation of the seismic system.

777222 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

Seismic Alarm and Monitoring System of INPP taken as a whole is one of the most advanced multifunctional seismic systems operated by large industrial plants worldwide. This is because it covers not only the usual industrial safety protection functionality but also the problems of seismic monitoring of surrounding areas. Therefore we believe that the upgraded system will make an important contribution to the seismic observations in Lithuania.

REFERENCES:

1. Peeters, B., G. Couvreur, O. Razinkov, C. Kündig, H. Van der Auweraer, G. De Roeck, Continuous Monitoring of the Øresund Bridge: System and Data Analysis, Proceedings of the International Modal Analysis Conference XX, January 2002 2. Erdik, M., Fahjan Y., Ozel O., Alcik H., Mert A., and Gul M., (2003b). Istanbul Earthquake Rapid Response and the Early Warning System. Bull. Of Earthquake Engineering, V.1, Issue 1, pp. 157–163 3. GeoSIG Newsletter GeoWatch, Issue 32, http://www.geosig.com/downloads/geowatch/GeoWatch32.pdf

777333 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

SSSEEEIIISSSMMMIIICCC MMMOOONNNIIITTTOOORRRIIINNNGGG OOOFFF AAANNN UUUNNNDDDEEERRRGGGRRROOOUUNNNDDD NNNUUUCCCLLLEEEAAARRR WWWAAASSSTTTEEE RRREEEPPPOOOSSSIIITTTOOORRRYYY AAATTT OOOLLLKKKIIILLLUUUOOOTTTOOO,,, FFFIIINNNLLLAAANNNDDD

Saari J. AF‐Enprima Ltd, POB 61, FIN‐16101 Vantaa, Finland, Rajatorpantie 8, Vantaa, GSM. +358 (0)40 348 5465, Fax. +358 (0)40 348 5007; [email protected]

The island of Olkiluoto, in the western coast of Finland, has been selected as the site for the final disposal facility of spent nuclear fuel. A Finnish expert organisation, Posiva started to construct an underground characterisation facility (the ONKALO) in Olkiluoto, in 2004. This facility will be used to acquire detailed information about the bedrock in Olkiluoto, to be utilised in the planning of the final disposal facility. The construction of the actual final disposal facility is scheduled to start in the 2015.The final disposal of spent nuclear fuel can be started in 2020. In February 2002, Posiva Oy established a local seismic network of six stations on the island of Olkiluoto. In the beginning of the year 2006 the seismic network consisted of eight seismic stations designed for monitoring tectonic and excavation induced seismicity inside the network. The main target volume of the seismic monitoring was the underground rock characterisation facility and the rockmass surrounding it. Monitoring of this volume is dominantly based on high frequency accelerometers. All the sensors were at the earth’s surface. First two borehole geophones were installed inside the ONKALO spiral at the end of 2006. Monitoring of semi‐regional tectonic seismicity and explosions started in February 2006. Four 1 Hz seismic stations locate from 3 to 7 km from the ONKALO. The fifth geophone is close to the ONKALO. In 2007, Posiva runs a seismic network of fourteen seismic stations. The microearthquake measurements in Olkiluoto aim for better understanding of the structure, behaviour and long‐term stability of the bedrock. The excavation works will change the characteristics (e.g. the stress field) of the virgin bedrock. Those changes will produce microearthquakes. The observed seismic events give an opportunity to approximate in what extent and where the bedrock is disturbed. Identification of active fracture zones is an essential element in a comprehensive study of potential hazards related to the spent nuclear fuel. The zones of weakness adjust releasing stresses and strains of the rock mass. The movements occurring on these zones accumulate during the lifespan of the repository and possibly can cause changes in the stability, stress field and groundwater conditions of the rock mass. The interpretation can bring out information that will be used for model calibration and even further cause changes to final disposal facility layout. An additional task of monitoring is related to safeguarding of the ONKALO. AF‐Enprima Ltd is responsible for the design and operation of the network as well as for data analysis and interpretation of the results. System is manufactured by ISS International Ltd. Generally, the sampling rate applied in the seismic stations is 6000 Hz (the ONKALO area) and 500 Hz (semi‐regional area). At every seismic station a data acquisition unit controls continuous data flow of the seismic sensors. When a pre‐set trigger value is exceeded a potential seismic signal is recorded to its hard disk drive. The Olkiluoto site central PC has an option to accept only events that trigger a certain pre‐set number of the sensors.

777444 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

When the event is detected, it is immediately emailed to the office PC in Vantaa, where it is automatically analysed. The location and magnitude of an event is determined when the email has arrived, basically in few minutes. The result of automatic analysis is uncertain and always verified manually. The decision of the seismic source (explosion or earthquake) is done by experienced analyst. This presentation introduce shortly nuclear waste management and general seismic setting in Finland as well as different applications related to seismic monitoring of underground nuclear waste repository.

777555 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

TTTHHHEEE RRROOOUUUTTTIIINNNEEE MMMIIICCCRRROOOEEEAAARRRTTTHHHQQQUUUAAAKKKEEE AAANNNAAALLLYYYSSIIISSS PPPAAACCCKKKAAAGGGEEE IIIMMMPPPLLLEEEMMMEEENNNTTTEEEDDD IIINNN IIICCCEEELLLAAANNNDDD (((IIIMMMOOO))) SSSNNNDDD SSSWWWEEEDDDEEENNN (((SSSNNNSSSNNN,,, UUUUUU)))

Slunga R. Swedish Defense Research Agency, FOI SE‐16490 Stockholm, Sweden, [email protected]

The Sedish Defense Research Institute (FOI) started Dec 1979 a seismic microearthquake network in Southern Sweden and Denmark for studies of the Swedish seismicity. The network consisted of 23 short period vertical seismometers covering an area of about 400 times 600 sqkms and with a station spacing of about 120 kms. The station had permanent analog connections to the central analog detectors and digitizers. Only when detections were achieved at 3 or more close stations within a reasonable time window were the data stored on digital tapes. So digital recordings were produced for time interval possibly containing an event. Most events were explosions but almost 1 Swedish EQ per day was recorded. All software for analyzing the digital recordings were developed at FOI. The software included digital phase detectors, automatic combination of the phase detections that could give locatable seismic sources and interactive analysis for improving the onset times and reading the first motion directions. The moment tensor inversion with double‐couple restriction (fault plane solution, FPS) was automaticly done based on the location and origin time of the EQ and on the interactively determined first motion directions. The method had been developed at FOI and was published in BSSA, Slunga 1981. It had turned out that high quality FPS could be achieved by use of the estimated low‐frequency spectral amplitudes (the time integrated displacements of the ground). Thus the routine analysis not only included the hypocenter, origin time, and the magnitude, but also the FPS, seismic moment, static stress drop, fault radius, and slip size. As the software is automatic large amounts of data can be handled without problem. When Ragnar Stefansson in started the SIL‐project in Iceland it was based on digital computer network systems developed by Reynir Bodvarsson in Uppsala and my microearthquake analysis software. The Icelandic microearthquake system started in 1990 and has today recorded over 250,000 microearthquakes in the range ‐10 are detected. The Swedish SNSN has a station spacing of about 70 km and is complete down to less than ML=0.5. The microearhquake software consists of four modules. The first, ANAAUT, makes multiphase event detection based on the stream of incoming phase detections from the stations. The detection results in a location, an origin time, and a preliminar magnitude plus a quality indicator. The second module is the interactiv software LOCIMP which allows the seismologist to improve the onset readings and add first motion directions. The third module is a signal analysis package, SKIAUT, which automaticly estimates the spectral amplitudes and corner frequencies needed for the mechanism analysis. The fourth module is the source mechanism software, SPQAUT, which makes the moment source tensor inversion based on the outputs of the previous packages. Only LOCIMP is not automatic all other steps can be run without use of LOCIMP and then automatic first motion directions are used.

777666 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

RRREEEAAASSSSSSEEESSSSSSMMMEEENNNTTT OOOFFF TTTHHHEEE DDDEEESSSIIIGGGNNN BBBAAASSSIIISSS EEAAARRRTTTHHHQQQUUUAAAKKKEEE FFFOOORRR IIIGGGNNNAAALLLIIINNNAAA NNNPPPPPP,,, NNNEEE LLLIIITTTHHHUUUAAANNNIIIAAA

Šliaupa S.1, 2, Kačianauskas R.3, Markauskas D.3 and Dundulis G.4 1Institute of Geology and Geography; [email protected] 2Vilnius University 3Vilnius Gediminas Technical University 4Lithuanian Energy Institute

The assessment of the seismic potential and related risk level of low seismicity areas is a highly complex problem. The Ignalina NPP, located in the East European Craton, was originally built for the minimum seismic risk conditions. Throughout the two decades of the operation of the plant several attempts were undertaken to reinforce the plant as the realization came of possibly higher seismic risk in the area. Still, the seismic potential of the Baltic basin is seemingly underestimated, as evidenced by the recent Kaliningrad earthquakes (2004, M = 5.0). It urges re‐evaluation of the seismic potential of the Baltic region in general and the local areas in particular where the high‐risk industrial objects are installed. The Design Basis Earthquake was re‐evaluated for the Ignalina NPP site. The Design Basis Earthquake of the site is estimated ML = 5.0 (Io= 7.5) and hypocentral depth of 10 km was assumed that is related to neotectonically active large‐scale Drūkšiai shear zone mapped close to the nuclear power plant. Here it intersects with some smaller scale faults. This tectonic feature is compatible to the other fault zones that show historical earthquake activity identified within the radius of 150 km from the plant. Only near‐field earthquake risk was considered, as the distant earthquakes are too weak to cause nay significant damages of the plant. In some previous studies the scaling of far‐field strong motion records (e.g. Carpathian earthquake, M = 8.5) was used in calculating the seismic load for the Ignalina NPP. It bears systematic error as dominant frequencies are shifted towards longer periods with increasing magnitude and distance. Therefore, the free‐field ground response spectra were calculated using attenuation relationship derived from Japanese near‐ field seismic records. The site‐specific amplification effects were taken into consideration. The rock and soil properties considerably influence the seismic wave propagating from the fault to the surface. Consistent data on the soil seismic properties of the Ignalina NPP site were collected during the seismic microzoning survey. S‐wave velocity measurements were performed in 20 shallow microzoning wells to the depth of 20 m. For the Unit 2 site, the average S‐wave velocity of the uppermost 0–10 m interval of the moraine dominated section is 270 m/s, increasing to 350 m/s in the 10–20 m depth interval. Following the ENV 1998‐1‐1 (Eurocode 8, 1998) classification, the soil of the Unit 2 site is attributed to the subsoil class B. Considering the underlying layers the S‐wave velocities were converted from P‐wave velocity measurements in the deep mapping wells drilled to the crystalline basement that occurs at the depth of 730 m. S‐wave velocities of the sedimentary rocks are in the range of 1190 m/s to 2400 m/s, while the crystalline basement is characterised by an S‐wave velocity 3150 m/s.

777777 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

The estimated design peak ground acceleration is as high as 0.166 g. It is higher than the minimum load (0.1 g) recommended by IAEA guidelines for SL‐2 ground motion hazard level. The maximum spectral acceleration is defined within 7‐10 Hz frequency range (Fig. 1). The same spectral shapes were assumed for two principle horizontal components N‐S and W‐E. It should be however noted that ground response spectra shape depends on the fault orientation. Assuming the earthquake generated at the Drūkšiai fault striking west‐east some reduction of the energy can be suggested for the horizontal component W‐E compared to the N‐S component in a low‐frequency range < 3Hz. This effect is still poor known and therefore was not accounted with in the Ignalina NPP spectra. The DynaTool program was used to obtain the time signals of the ground acceleration (synthetic accelerograms) of the Ignalina NPP site (Fig. 1). The time‐histories were derived following guidelines by ASCE 4‐95 and IAE NS‐G‐1.6, the enveloping function was accordingly assumed 1+6+3 s. The backward solution indicates consistency of the method applied The in‐structure response spectra were calculated for different levels of the Unit 2 Rector Building. They differ considerably from previous estimates by higher load in the high‐ frequency range, whereas much lower values are estimated for the low‐frequency range.

Fig. 1. Free‐field ground response spectra (5% damping, horizontal and vertical components) and horizontal time‐history of the Ignalina NPP Design Basis Earthquake

777888 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

NNNOOORRRWWWEEEGGGIIIAAANNN NNNAAATTTIIIOOONNNAAALLL SSSEEEIIISSSMMMIIICCC NNNEEETTTWWWOOORRRKK RRREEEAAALLL TTTIIIMMMEEE MMMOOONNNIIITTTOOORRRIIINNNGGG

Utheim T., Havskov J. Department of Earth Science, University of Bergen, Allegaten 41, 5007 Bergen, Norway, [email protected], [email protected]

The Norwegian National Seismic Network (NNSN) consists of around 30 stations, where nearly all have Internet communication to the data center in Bergen. Data, both events and continuous, are downloaded to the data center at regular intervals for further processing, on average up to 12 hours after real time using the UiB SEISNET software. Due to relatively high cost of Internet over telephone lines, continuous data transmission has not been an option. Access to low cost Internet on most field stations, standardization of data formats and new software now makes near real time data transmission possible and economical. UiB SEISLOG field stations (Windows or Linux) and SEISLOG embedded systems, now transmit data in a well defined standard format (LISS MiniSeed blocks) which can be read by systems like SeisComp/SeedLink and Earthworm. An NNSN field station typically consists of a PC, a digitizer and sensor where the best are STS2 with EarthData digitizer (EU 20000) and the simplest are SARA digitzers with 4.5 Hz geophones (EU 2000). Testing is now being done at UiB running a real time system in parallel with the SEISNET system. A central SeedLink server reads data in real time from several field stations. The stations are a mix of SEISLOG Windows, SEISLOG Linux, SEISLOG Embedded and also stations in the GSN network which are transmitted through the LISS server in Albuquerque. Stations from the UK and Finland are also included and a total of 18 stations now enter the real time system. The SeedLink server serves as storage of continuous data as well as a provider of data for several monitoring clients and for Earthworm. Earthworm on Sun Solaris reads data from the SeedLink server and do event detection. Events are stored in the standard NNSN SEISAN database for processing. As SeedLink makes it relatively easy for user‐written clients to access data in real time, we have also written our own program RTDET for detecting events in the real time data stream. Event data are extracted from the SeedLink server directly and written in the standard SEISAN database. The detection client may run on a local machine or from anywhere if connected to Internet. The intention with RTDET is to develop a simple alternative to the more complex Earthworm. We have also made some utility clients to monitor the data on the SeedLink server in near real time (RTNET,RTMON). The experience with the LISS protocol has been good even if there is no re‐transmission integrated. An improvement is to install a SeedLink server on the field station where possible. Then LISS data are only transmitted internally on the field machine. Earthworm triggers have been compared to SEISNET triggers and the conclusion is that Eartworm works well. The plan is to get most stations operating in real time (with error correction) in the near future. All UiB software is public domain: www.geo.uib.no/seismo Seiscomp is obtained from: ftp://ftp.gfz‐potsdam.de/pub/home/st/GEOFON/software/SeisComP/2.5/ Earthworm is obtained from: http://folkworm.ceri.memphis.edu/ew‐dist/ Earthdata from: www.earthdata.co.uk SARA systems from: www.sara.pg.it

777999 AND SEISMOLOGICAL OBSERVATIONS OF THE BALTIC SEA REGION SEISMICITY VVVOOOLLLUUUMMMEEE OOOFFF AAABBBSSSTTTRRRAAACCCTTTSSS SEISMICITY AND ADJACENT TERRITORIES September 10–12, 2007, Lithuania

TTTHHHEEE SSSEEEIIISSSMMMOOOLLLOOOGGGIIICCCAAALLL NNNEEETTTWWWOOORRRKKK IIINNN DDDEEENNNMMMAARRRKKK AAANNNDDD IIINNN GGGRRREEEEEENNNLLLAAANNNDDD,,, EEEAAARRRTTTHHHQQQUUUAAAKKKEEE MMMOOONNNIIITTTOOORRRIIINNNGGG AAANNNDDD AAAPPPPPPLLLIIIEEEDDD RRREEESSSEEEAAARRRCCCHHH

Voss P. Geological Survey of Denmark and Greenland, Oster Voldgade 10, DK‐1350 Copenhagen K, Denmark; [email protected]

The Geological Survey of Denmark and Greenland (GEUS) operate the national seismological network in Denmark and in Greenland. In Denmark the network consists of 3 broad band and 2 short period seismometers, all permanent. In Greenland the network consists of 17 broad band seismometers, 4 are permanent and 13 are temporary. The data from the stations in Denmark are collected through the Internet. Data from two stations in Greenland are collected through the Internet, from the rest of the stations data are either mailed weekly on tapes or flash cards or collected at a yearly inspection. The data is stored and processed in SEISAN databases, which also contain the GEUS earthquake catalogues for Denmark and Greenland. The seismological networks are currently part of four research projects: (1) A UN article 76 based study of crustal structures in North Greenland. (2) A study of lithospheric structures in West Greenland initiated by the diamond exploration. (3) A study of glacial earthquakes in East Greenland. (4) PASSEQ – a teleseismic tomographic study of the lithospheric structures in Germany, Poland and Lithuania. These and previous research projects have improved our knowledge and understanding of the development of the earth, contributed to the education of several university students, and furthermore, the applied research ensures that the quality of the network is kept at a high standard. The status of the seismological network in Denmark and in Greenland, the earthquake monitoring and the current research projects are presented.

888000