Bursa ve çevresindeki küçük depremlerin ivme kayıtlarının incelenmesi

Abstract

Bu çalışmada, Bursa ve çevresine yerleştirilmiş 5 adet ivme-ölçer (strong-motion) sismografından elde edilen küçük depremlerin ivme ve yerdeğiştirme spektrumları kullanılarak kaynak parametrelerinin belirlenmesine çalışılmıştır. Bu amaçla Kasım 1992 - Şubat 1994 arasında olmuş ve 28.8° - 29.5° D boylamları ile 39.9° - 40.5° K enlemleri arasında kalan 40 depremin 3 bileşenli sayısal ivme sismogramı incelenmiştir. Spektral özelliklerin ve kaynak parametrelerinin belirlenmesi amacıyla 40 ivme sismogrammın SH (transvers S) ivme ve yerdeğiştirme spektrumları hesaplanmış, köşe frekansı fQ ile i"max parametreleri incelenmiştir. Uzak alan Brune kaynak modeli kuramından yola çıkarak spektral parametreler belirlenmiş; bu parametreler kullanılarak kaynak boyutu, sismik moment, gerilme düşümü ve sismik enerji gibi kaynak parametreleri elde edilmiştir. Nedeni ve fiziksel mekanizması tartışmalı olan fmax parametresi incelenmiş ve bu parametrenin kaynak ile ilgili olduğu sonucuna varılmıştır. Tortul kütleler üzerinde bulunan istasyonlardan elde edilen SH spektrumlarmda çeşitli Qs etkileri incelenmiştir. Köşe frekansı f' da az da olsa değişim gözlenmiş, f ' m seçiminin ise güçleştiği görülmüştür. Kaya ve temel birimlerde soğurulmanm az, spektral parametrelerde ise belirgin değişimin olmadığı sonucuna varılmıştır. Yerdeğiştirme verilerine geçilmeye gerek kalmaksızın ivme spektrumundan köşe frekansının okunabileceği ve buna bağlı kaynak parametrelerinin hesaplanabileceği bulunmuştur. Elde edilen dinamik ve kinematik kaynak parametreleri arasmda bağıntılar elde edilmiş ve tartışılmıştır.

In this study, the dynamic and kinematic source parameters of small earthquakes which were recorded by BURSA ACCELERATION NETWORK (BUSNET), were determined by using SH spectra. The network consists of 5 tree component digital seismographs deployed in and around Bursa city. Bursa city geographically located at south of Marmara region. Marmara region is a part of western Pontids which are made up of three major tectonic units developed in Mid-to-Late Mesozoic times. These are Istranca Massif in the west, Istanbul Zone further east and Sakarya Zone in the southwest. Geologically, Bursa region is a part of the Sakarya Zone in western Pontids. The Sakarya Zone is characterized by a variably metamorphosed and strongly deformed Triassic basement called the Karakaya Complex. The base of the Karakaya Complex is exposed in the Uludağ and Kazdağ tectonic windows in Western Anatolia where high-grade gneiss, marbles, amphibolites and metaperidotites occur tectonically beneath the low- grade metabasites of the Karakaya Complex. Western Türkiye which is called Western Anatolia Province [11] is one of the neotectonic provinces of Türkiye and is dominated by north-south extension. The noetectonic period of Western Anatolia Extensional Province which consists of Marmara Region and Southwestern Anatolia is originated following the collusion of Arabian and Anatolian plates during the middle Miocene. The western end of the North Anatolian Fault (NAF) zone bifurcates into several branches in the Marmara region. The seismotectonics characteristics of the Marmara region dominated by the interaction between north-south extension of western Türkiye and right-lateral displacements along xni to western branches of NAF zone. The strong-motion seismographs were deployed on various geologic units in around Bursa. The acceleration recorders were installed mainly on the sites where thick Permian limestone (solid rock), and thick Noegene formations (stiff soil or sandstones) exposed. The Demirtaş station (SDEM) installed on the Neogene sandstones. The İğdır station (SIGD) located at the İğdır village, the northeast of Bursa city. The station was installed on the thick Permian limestones. The Hamamhkizik station (SHMK) at the east of Bursa region was installed on the slope deposits of Uludağ Mountain. The Kaphkaya station (SKAY) is deployed on the foothills of Uludağ Mountain where the crystallized limestone exposed. The Hamitler station (SHMT) is at the west of Bursa region. The station was installed on the thick Neogene formations. Three major earthquake catalogues are merged in order to investigate the seismicity of Bursa and its vicinity in a wide range of time interval. These catalogues are Ergin [22], Soysal [23] and ISC (1964-1987) catalogues. For the historical earthquakes (pre- 1900) Soysal's catalogue has been used (Table 2.1). This catalogue covers the time period from 2000 BC to 1900 AD. The earthquakes occurred during the instrumental period (post-1900) were selected from two latter catalogues. Ergin [22]' s catalogue was used for the earthquakes occurred in the period of 1900-1963. The earthquakes occurred between the time period of 1964 and 1990 have been compiled from ISC catalogue. These three earthquake catalogues have been reformatted and unified in order to obtain a single catalogue for the time period between 2000 BC and 1990 AD. It is seen from the epicentral map that the seismicity of Bursa region is high during the historical and instrumental period and displays swarm type earthquake activity. In recent years numerous studies have been done concerning source parameters and spectral scaling of small to large earthquakes. The popular source models widely used by seismologist are Haskell [l]'s, Aki [2]'s and Brune [3]'s source models. The earthquakes source model (fault model) can be classified into two types : a "kinematic model" and a "dynamic model". The kinematic model is based on elastic theory of dislocation and also called a "dislocation model". A dislocation model requires the temporal and spatial distribution of the dislocation at both side of crack (fault). The dynamic model is based on fracture mechanics. In this method, the fracture criteria of the fault and the condition of the stress are given. Haskell [1] model is xiv the most popular model among the kinematic models. He described the fault model by five source parameters; the fault length L, the fault width W, the average dislocation D, rise time x and rupture velocity V. Haskell [1] recognized that the assumption of coherent radiation, implicit in his theory, would break down at sufficiently high frequency because of the inherent irregularity of the slip process during the earthquake. By using the kinematic model concept, Haskell [1] and Aki [2] proposed w-square and CD-cube models. Aid [2] developed «-square statistical model further. Brune [3] proposed a relatively simple model of faulting that affords tractable expressions for calculating some of the parameters of faulting. Three independent spectra parameters are obtained from Brune [3]'s source spectrum. These are long period spectral level 6)Q, which is proportional to seismic moment M0; the spectral corner frequency fQ, which is proportional to the reciprocal of the radius are of a circular fault area; and a parameter e, which controls the high frequency (f>fQ) decay of the displacement spectrum. Savage [4] discussed differences of source models developed by Brune [3] and Haskell [1]. Both theories provided for a spectrum in which the average trend in the middle frequencies is proportional to co, but the cause of this intermediate trend is entirely different in the two models. Based on the statistical source model developed by Aki [2] it is suggested that the corner frequencies at the spectrum of the to -square model depend on the coherence length kL" and the coherence time LjT rather than on the fault length L and the characteristic time T. The implications that for the larger earthquakes the corner frequencies determine coherence length and coherence time rather than fault length and fault width. These results show that study of the source characteristics of the small earthquake may give important clues about the relations between source size and temporal development of the rupture in a small scale earthquake which can be simulated to large earthquakes. Based on the present source model, it is generally believed that the spectral displacements beyond f decrease at a rate proportional to to- squared. Hanks [7] and Papageorgiou [51] discussed the existence of a second corner frequency, namely f, beyond which the decay of frequencies depart from a> -square and o -cube models. fmax was attributed by Hanks [7] to the effect of absorption due to the near surface effect of the earth. Originally, Papageorgiou [51] attributed fmax to the source effect, without considering the site effect. Later, Papageorgiou [52] showed that fmax was not significantly different between a soil site and a rock site in the case of the San Fernando earthquake. A lot of scientist worked on fmax till today, but the origin of fmax is not proved absolutely. xv In this study, the 40 three-component digital accelograms of small earthquakes obtained at distances less than 100 km in Bursa area (39.9 - 40.5° N latitude and 28.8 - 29.5° E longitudes) were analyzed in order to determine spectral source parameters. Arrival times of P and S waves were collected from BUSNET stations and from other local stations operated by TÜBİTAK Marmara Research Center and Boğaziçi University Kandilli Observatory in order to locate the earthquakes. The earthquakes are relocated for a layered crustal structure. Using back-azimuths, the ground NS and EW components were rotated into SH and SV components. SH acceleration seismograms were integrated twice to get the displacement records. The resulting time series were multiplied 10 percent cosine window and FFT was applied to calculate both acceleration and displacement spectra. The appearance and the decay rate of the frequencies beyond fQ for many source spectra depart from Brune [3]'s source model. One of the aim of this study is to show the possibility of obtaining the corner frequency f from the acceleration record without converting the signal to displacement record. SH acceleration and displacement spectra were carefully examined to find spectral parameters g>0 and fQ. From these spectral parameters the seismic moment, source size, stress-drop, energy and effective stress were calculated. The spectra were also examined to find out the existence of the cut-off frequency fmax beyond which the spectral amplitudes decay sharply. The validity of corner frequency fQ and fmax tested in the domain of both acceleration spectra an displacement spectra. The attenuation effect was removed from most of the observed SH spectra. The effect of various Qs models on the spectral parameters is also discussed. It is observed that fmax was not changed drastically after the correction of Qs effect. This implies that fmax can be attributed as a source-controlled parameter. Nevertheless, the relation between fmax and magnitude cannot be found because of very small magnitude range involved. Duration magnitudes (Md) of earthquakes studied range from 1.7 to 3.5. The source parameters are calculated using the formulas inferred from Brune's source model, however some of the spectra depart from that model. Seismic moments range from 10 to 10 dyne-cm. The SH-wave energies 13 15 of earthquakes vary between 10 - 10 ergs. Stress drops range from 1 to 10 bars. Corner frequencies vary between 2 - 13 Hz at the acceleration spectra while they range from 2 to 10 Hz at the displacement spectra. The radii of the earthquake sources obtained between the values of 90 and 560 m from the acceleration spectra while they obtained between the values of 120 and 520 m from the displacement spectra. xvi The relation between the spectral source parameters which are calculated from acceleration and displacement spectra were given below :. Magnitude fMd) - Seismic Moment (M0) For the displacement data : log MQ = 0.71 Md + 18.42 (r = 0.69). Seismic Moment (MQ) - Stress drop (A a) For the acceleration data : log A a = 0.38 log M0 - 7.14 (r = 0.55) For the displacement data : log A a = 0.80 log MQ - 15.57 (r = 0.65). Magnitude (Md) - Source dimension (r) For the acceleration data : log r = 0.17 Md - 1.02 (r = 0.55) For the displacement data : log r = 0.04 Md - 0.66 (r = 0.11). Magnitude (Md) - Seismic Energy (E5) For the acceleration data : log Es = 0.93 Md + 12.25 (r = 0.59) For the displacement data : log Es = 1.36 Md + 11.12 (r = 0.66). Here, r is the correlation coefficient.