##
AVO analizi ile deniz tabanının modellenmesi

AVO analizi ile deniz tabanının modellenmesi

##### Dosyalar

##### Tarih

1997

##### Yazarlar

Ocakoğlu, Neslihan

##### Süreli Yayın başlığı

##### Süreli Yayın ISSN

##### Cilt Başlığı

##### Yayınevi

Fen Bilimleri Enstitüsü

##### Özet

Bu çalışmada deniz tabanının Yansıma Genliği-Açılım Analizi (Amplitude-versus- offset veya kısaca AVO) ile modellenmesi gerçekleştirilmiştir. Modellemede Saroz Körfezi'nden alınan bir grup CDP verisine Deneme- Yanılma (Trial and Error) yöntemi ile düz çözüm uygulanarak deniz tabanının elastik parametreleri hakkında bilgi edinilmiştir. Bir ortamda yayılan ve bir arayüzeyden yansıyan sismik dalgaların, ortamların litolojik özelliklerine işaret edebilecek şekilde yansıma genliğini etkileyen faktörler; arayüzeyin alt ve üstünde yeralan ortamların P ve S dalga yayınım hızları, yoğunlukları ve dalgaların arayüzeye geliş açılandır. Dalgaların arayüzeye geliş açısı bir değişken olarak kontrol edilebildiğinden yansıyan dalgaların genliğinin geliş açısı ile değişimi ortamların fiziksel özellikleri hakkında bilgi sahibi olmamıza olanak verir. Sismik dalgaların arayüzeye geliş açılan aslında sismik enerji kaynağı ve alıcılar arasındaki uzaklık (offset) ile kontrol edilebilir. Bu çalışmada yansıma genliklerinin açılıma bağlı değişimleri incelenmiştir. AVO veri-işleminde bilinen sismik veri-işlemden farklı olarak göreceli genlik değerleri (Relative Amplitude) önemlidir yani AGC (Automatic-Gain -Control) uygulanmamış sismik veriler kullanılır. Uygulamada akustik/elastik bir ortam olan su/kaya ortamında deniz tabanından yansıyarak gelen dalgaların genliklerini modellemek için Zoeppritz denklemleri kullanılmıştır. Düz çözüm aşamasında Saroz körfezinden alınan CDP (Common- Depth-Point, ortak derinlik noktası) Açılım- Yansıma Genliği verileri programda giriş verisi olarak kullanılmıştır. Modelleme öncesi CDP verileri S/G oranını arttırmak için ayıklanmıştır. Bu aşamada CDP verisi üç gruba ayrılarak her bir grup kendi içinde düşey yığma (Vertical Stack) işlemine tabi tutulmuş daha sonra her gruba ait açılıma bağlı yansıma genlikleri öncelikle Q soğurulma düzeltmesi uygulamasından geçirilmiş ve daha sonrada genlikler okunmuştur. Okunan genlik değerlerine programda önce kayan ortalama yuvarlatma işleci ve küresel yayılma düzeltmesi (Geometrical Spreading) uygulanmış sonraki adımda da seçilen fiziksel kıstasların ışığında Deneme- Yanılma tekniği ile model veri üretilmiş, üretilen model veri ile gözlemsel veri kıyaslanarak matematik koşullan sağlayan model verinin parametreleri saklanmıştır. Böylece ortamı en iyi temsil eden P, S dalga hızları ve yoğunluk parametrelerinin frekans histogram grafikleri elde edilmiştir. Elde edilen sonuçlara göre; %90 ihtimalle deniz tabanı siltli-killi birimler için hesaplanan P dalga hızı, S dalga hızı ve ortamın yoğunluğu şöyledir: P dalga hızı Vp = 1575 m/sn, S dalga hızı Vs = 850 m/sn, yoğunluk p = 1.7 gr/cm3. Aynca deniz tabanı için hesaplanan R(0) değeri yani t=0 düşey gidiş geliş zamanındaki yansıma genliği ve Poisson değeri yaklaşık olarak şöyledir: R(0) = 0.285, o =0.28. vııı

In Reflection Seismology, the estimation of P and S wave seismic velocities and densities of Earth layers has long been one of the main research subjects. The amplitude of reflected seismic wave is dependent upon the incidence angle and the P and S wave velocities and densities of two media seperated by a flat discontinuity. In fact, incidence angle of seismic waves arrived to the boundary can be related to offset, the distance between source and receiver. Therefore, the variation of reflection and transmission coefficients with increasing offset is referred to as offset-dependent-reflectivity and is the fundamental basis for amplitude-versus- offset (AVO) analysis. AVO analysis gives us knowledge about physical properties of Earth's crust. This knowledge will lead lithological interpretations. Knowledge of lithology will open the ways to direct hydrocarbon exploration, ground water exploration and understanding of the Earth's crust. The reflection, transmission and mode conversion of the plane P waves at a boundary as a function of incidence angle have been extensively investigated in the geophysical literature. Seismic waves propagating in the Earth and reflected and transmitted from interfaces. The mathematical description of this energy partitioning at a boundary is given by the Zoeppritz equations (Waters, 1987; Berkhout, 1987). Curves of the P wave reflection coefficients obtained from the Zoeppritz equations were developed for different models by several authors, Koefoed (1955), Bortfeld (1961), Shuey (1985), Backus (1987). Forward and inverse modelling methods were developed to make interpretations for the seismic parameters in the reflection process by Rutherfold and Williams (1989), Keys (1989), Silva and Ahmed (1989), Demirbağ at all, (1993), Richard (1989). For example; Rutherford and Williams (1989) classified the gas sands on the basis of their acoustic impedances and AVO responses for a shale-gas sand model, and gave examples of field data for each class. Demirbağ at all, (1993) studied a multilayer inversion method GLI (Generalized Linear Inversion ) to estimate the P and S wave velocities and densities from AVO data CMP (common-mid-point). Silva and Ahmed (1989) used AVO inversion technique as an interpretation tool for pore-fluid boundaries/contacts in hydrocarbon potential formations. IX Forward modelling is used to determine the response of a physical system using mathematical relations. In this study, a forward method which is called Trial and Error is developed to estimate the P and S wave velocities and density of an acoustic/elastic media separated by a flat, horizontal boundary. The Zoeppritz equations (Berkhout, 1987) for an fluid/solid media as a physical model, AVO data from CDP (common-depth-point) gathers as input were used and distribution functions of the estimated seismic parameters as output were calculated. Using the continuity of displacements and stresses at the boundary for a plane P-wave incidence with unit amplitude, the Zoeppritz equations are obtained as a system of equations in matrix form (Berkhout, 1987). To calculate the P wave reflection amplitude coefficients; Snell's law, mathematical trigonometric equations and model geometry (CDP geometry) were used and incident angle was transformed into offset domain. Then, the variation of P wave reflection amplitudes was calculated with increasing offset. All of these approximations apply only to pre-critical angles of incidence. The reflection amplitude has a maximum at the critical incidence. For normal incidence, the reflection amplitude depends on the P wave velocity and density of the upper and lower media and independent of S wave velocity. In application, calculation of P wave reflection coefficients for a water/rock media provides an advantage of simplifying Zoeppritz equations. Because, upper media P wave velocity (1500 m/sec) and water density (1.0 g/cm3 ) are known. All reflection amplitudes are normalized by zero-offset reflection amplitude. In this study, the chosen model is an water/rock media with a single horizontal interface. Known seismic parameters are upper medium P wave velocity and density. The target is to calculate the distribution function of lower media seismic parameters (P, S wave velocities and density) by Trial and Error method. In this modelling, a group of shallow marine seismic CDP gathers from Saroz Bay area was used. CDP gather is the most suitable form of data to observe the variation of reflection amplitude with offset. San, Özel ve Ergün (1989) researched tectonics and sedimentary evolution of the Saroz Bay area. The basins of the North Aegean and the Marmara Sea show the characteristics of rapid subsidence accompanied by extension and transform motion. In the Saroz Bay area pure strike-slip motion changes into extensional strike-slip movement responsible for the creation of basins of the Marmara Sea and the Saroz Bay where the Ganos Mountain active faults joins these two basins. The evolution of these basins is complicated because they have suffered several compressional and extensional tectonic regimes which have resulted in superimposed deformations. The general framework of this region is relatively young, and superimposed on older structures of an orogenic belt of Alpidic origin. Three sedimentary sequences can be described from the seismic reflection data as: (i) Upper-Lower Eocene sequence; (ii) Middle Eocene-Oligocene sequence; (iii) Mio-Pliocene-Quaternary sequence. In modelling, for estimations of initial model parameters of the Saroz Bay, Sari's research (unpublished master thesis, 1997) was used. In his study, geochemical investigation of the Sea floor sediments of Saroz Bay and distribution map of sediments are considered. P wave velocity and density of sedimentary rocks are smaller than those of the magmatic and metamorphic rocks. Therefore calculated zero offset reflection amplitudes R(0) for sedimentary rocks generally varies between (-0.5, +0.5). In this study, AVO data (CDP gathers) is corrected before forward modelling. Anelastic attenuation correction and spherical divergence correction are applied to observed data. Transmission effects and multiples are omitted because we are interested in only the primary reflections of the sea floor. Anelastic attenuation is the loss of energy of seismic waves due to the conversion of wave energy into heat energy by internal friction of the material in which the seismic waves propagate. For a constant velocity medium and a given distance, the high frequencies are attenuated faster than the low frequencies because there are a greater number of wavelengths of high frequencies than low frequencies. The loss of energy or decrease in amplitude of a sinusoid is defined in variety of ways e.g. quality factor Q. For anelastic attenuation correction, Çoruh et al., (1990) equation was used. To estimate Q quality factor, Bourbie, Olivier e.g. (1987) applications for Q factor estimation of different rocks were considered. In AVO analysis relative reflection amplitudes are modeled. That is, there is no AGC (automatic gain control) applied on the data. In this study, determination of the offset dependent reflection amplitudes has been one of most the critical stage. First of all, all CDP gathers are divided by three CDP gathers group. Then, the vertical stack is applied to every CDP gathers group to increase S/N (signal/noise) ratio. So, with a chosen cosine windowing, Peak-to-Trough method is applied to data to measure amplitude of the sea bottom reflected wavelets from trace to trace on every CDP gathers group. In input data, all amplitudes were chosen positive polarity. In second step, all determined offset dependent amplitudes are smoothed by moving average operator then, spherical divergence correction is applied to the three CDP gather groups. The decrease in the amplitude of seismic waves due to geometrical conditions is known as spherical divergence and it is discussed in detail in the literature, e.g., O'Doherty and Anstey (1971) and Sheriff (1975). Seismic waves generated at a point source propagate as spherical wavefronts in a homogeneous and isotropic medium. There is a simple relation between increased distance and decreased energy. The energy is decreased as the squared distance increases. Because the amplitude is defined as the square root of the energy, amplitude decreases as the distance increases. Although this simple relation is true for a homogenous, isotropic Earth model, most of the time the velocity change with depth in a layered Earth. In this study, a solution for the divergence problem at an sea/rock media which upper and lower media are homogeneous and isotropic, Newman's equation (1973) was used. All stages given above are preprocess studies. After spherical divergence correction, forward modelling with Trial and Error method is applied to observed data. In forward modelling, input includes central model parameters (lower layer P and S wave velocities and lower layer density), user defined limits (for P, S wave velocities, XI density end Poisson ratio), number of Trial and Error, fit tolerance, number of interfaces, zero offset reflection time, upper media P wave velocity and P wave reflection amplitudes with increasing offset. To estimate density limits of the lower media, Gardner's relation (1974) was used. In determination of Poisson limits, the relation between Vp/Vs and Poisson value for sedimentary rocks is examined. The sea floor of studied area of Saroz Bay is composed of silt and clay. This knowledge gives an important advantage of determination central model parameters. In first step, the model parameters (lower media P and S wave velocities and density) were produced as random numbers. If the physical conditions are satisfied (these are user defined limits), the model parameters are sent to Zoeppritz equations to calculate P wave reflection amplitudes coefficients. The maximum offset was 1450 m. long and the boundary was located at 675 m depth. The model AVO data fall in the precritical region of offset with the maximum angle of incidence of 47° (the critical angle is about 56°). Calculated model AVO data response is normalized to zero offset reflection amplitude. Then, the observed data and model data are compared. If the mathematical constraints satisfied the model parameters are stored. To compare model and observed data, fit tolerance factor is considered. That is, if the produced model parameters can be approached to the observed AVO data with respect to the fit tolerance factor, these parameters are stored. Else, new model parameters are produced and all the stages above are repeated. This trial repeats until the number of Trial and Error. When the number of stored model parameters reached five hundreds, the distribution of solutions for each seismic parameters is calculated. The solutions are normalized. For each distribution function of seismic parameters (lower media P wave velocity, lower media S wave velocity and density), three values are calculated and desired the most likely solutions for each parameter are accepted with 90% confidence intervals among the three values. In the last step, AVO model data (P wave reflection amplitudes coefficients) is calculated with respect to the most likely solutions in the Zoeppritz equations. Then, R(0) zero offset reflection amplitude, error which is the square of between observed and model data differences and Poisson value of lower media are calculated. The results and comments are as follows : * In AVO analysis, considered amplitudes are relative amplitudes. That is, there is no AGC on AVO data. * The basis of AVO analysis is to determine the variation reflection amplitudes with offset. The variation of reflection amplitudes is small in the near offsets but high in the far offsets. Especially, near precritical angle offsets, it takes maximum value. xn * In application, the trend in the AVO data is easily noticeable due to high S/N ratio. As the S/N ratio is increased, the resolving power of the forward method increases and the estimated results approach to the model parameters. Before modelling, to increase the S/N ratio, vertical stack is applied to CDP data. Then, AVO data is smoothed by an operator. * In application, with Trial and Error method, forward modelling is applied to AVO data, and the seismic parameters of Saroz bay sea floor are calculated. Sea floor is composed of unconsolidated silt and clay materials. P wave velocity of these materials is approximately 1575 m/sec., S wave velocity is 850 m/sec., the density is 1.7 g/cm3. The estimated parameters are suitable to the searched values for sedimentary rocks. Adding to this, the calculated R(0) is 0.285 and Poisson value is approximately 0.28. xiu

In Reflection Seismology, the estimation of P and S wave seismic velocities and densities of Earth layers has long been one of the main research subjects. The amplitude of reflected seismic wave is dependent upon the incidence angle and the P and S wave velocities and densities of two media seperated by a flat discontinuity. In fact, incidence angle of seismic waves arrived to the boundary can be related to offset, the distance between source and receiver. Therefore, the variation of reflection and transmission coefficients with increasing offset is referred to as offset-dependent-reflectivity and is the fundamental basis for amplitude-versus- offset (AVO) analysis. AVO analysis gives us knowledge about physical properties of Earth's crust. This knowledge will lead lithological interpretations. Knowledge of lithology will open the ways to direct hydrocarbon exploration, ground water exploration and understanding of the Earth's crust. The reflection, transmission and mode conversion of the plane P waves at a boundary as a function of incidence angle have been extensively investigated in the geophysical literature. Seismic waves propagating in the Earth and reflected and transmitted from interfaces. The mathematical description of this energy partitioning at a boundary is given by the Zoeppritz equations (Waters, 1987; Berkhout, 1987). Curves of the P wave reflection coefficients obtained from the Zoeppritz equations were developed for different models by several authors, Koefoed (1955), Bortfeld (1961), Shuey (1985), Backus (1987). Forward and inverse modelling methods were developed to make interpretations for the seismic parameters in the reflection process by Rutherfold and Williams (1989), Keys (1989), Silva and Ahmed (1989), Demirbağ at all, (1993), Richard (1989). For example; Rutherford and Williams (1989) classified the gas sands on the basis of their acoustic impedances and AVO responses for a shale-gas sand model, and gave examples of field data for each class. Demirbağ at all, (1993) studied a multilayer inversion method GLI (Generalized Linear Inversion ) to estimate the P and S wave velocities and densities from AVO data CMP (common-mid-point). Silva and Ahmed (1989) used AVO inversion technique as an interpretation tool for pore-fluid boundaries/contacts in hydrocarbon potential formations. IX Forward modelling is used to determine the response of a physical system using mathematical relations. In this study, a forward method which is called Trial and Error is developed to estimate the P and S wave velocities and density of an acoustic/elastic media separated by a flat, horizontal boundary. The Zoeppritz equations (Berkhout, 1987) for an fluid/solid media as a physical model, AVO data from CDP (common-depth-point) gathers as input were used and distribution functions of the estimated seismic parameters as output were calculated. Using the continuity of displacements and stresses at the boundary for a plane P-wave incidence with unit amplitude, the Zoeppritz equations are obtained as a system of equations in matrix form (Berkhout, 1987). To calculate the P wave reflection amplitude coefficients; Snell's law, mathematical trigonometric equations and model geometry (CDP geometry) were used and incident angle was transformed into offset domain. Then, the variation of P wave reflection amplitudes was calculated with increasing offset. All of these approximations apply only to pre-critical angles of incidence. The reflection amplitude has a maximum at the critical incidence. For normal incidence, the reflection amplitude depends on the P wave velocity and density of the upper and lower media and independent of S wave velocity. In application, calculation of P wave reflection coefficients for a water/rock media provides an advantage of simplifying Zoeppritz equations. Because, upper media P wave velocity (1500 m/sec) and water density (1.0 g/cm3 ) are known. All reflection amplitudes are normalized by zero-offset reflection amplitude. In this study, the chosen model is an water/rock media with a single horizontal interface. Known seismic parameters are upper medium P wave velocity and density. The target is to calculate the distribution function of lower media seismic parameters (P, S wave velocities and density) by Trial and Error method. In this modelling, a group of shallow marine seismic CDP gathers from Saroz Bay area was used. CDP gather is the most suitable form of data to observe the variation of reflection amplitude with offset. San, Özel ve Ergün (1989) researched tectonics and sedimentary evolution of the Saroz Bay area. The basins of the North Aegean and the Marmara Sea show the characteristics of rapid subsidence accompanied by extension and transform motion. In the Saroz Bay area pure strike-slip motion changes into extensional strike-slip movement responsible for the creation of basins of the Marmara Sea and the Saroz Bay where the Ganos Mountain active faults joins these two basins. The evolution of these basins is complicated because they have suffered several compressional and extensional tectonic regimes which have resulted in superimposed deformations. The general framework of this region is relatively young, and superimposed on older structures of an orogenic belt of Alpidic origin. Three sedimentary sequences can be described from the seismic reflection data as: (i) Upper-Lower Eocene sequence; (ii) Middle Eocene-Oligocene sequence; (iii) Mio-Pliocene-Quaternary sequence. In modelling, for estimations of initial model parameters of the Saroz Bay, Sari's research (unpublished master thesis, 1997) was used. In his study, geochemical investigation of the Sea floor sediments of Saroz Bay and distribution map of sediments are considered. P wave velocity and density of sedimentary rocks are smaller than those of the magmatic and metamorphic rocks. Therefore calculated zero offset reflection amplitudes R(0) for sedimentary rocks generally varies between (-0.5, +0.5). In this study, AVO data (CDP gathers) is corrected before forward modelling. Anelastic attenuation correction and spherical divergence correction are applied to observed data. Transmission effects and multiples are omitted because we are interested in only the primary reflections of the sea floor. Anelastic attenuation is the loss of energy of seismic waves due to the conversion of wave energy into heat energy by internal friction of the material in which the seismic waves propagate. For a constant velocity medium and a given distance, the high frequencies are attenuated faster than the low frequencies because there are a greater number of wavelengths of high frequencies than low frequencies. The loss of energy or decrease in amplitude of a sinusoid is defined in variety of ways e.g. quality factor Q. For anelastic attenuation correction, Çoruh et al., (1990) equation was used. To estimate Q quality factor, Bourbie, Olivier e.g. (1987) applications for Q factor estimation of different rocks were considered. In AVO analysis relative reflection amplitudes are modeled. That is, there is no AGC (automatic gain control) applied on the data. In this study, determination of the offset dependent reflection amplitudes has been one of most the critical stage. First of all, all CDP gathers are divided by three CDP gathers group. Then, the vertical stack is applied to every CDP gathers group to increase S/N (signal/noise) ratio. So, with a chosen cosine windowing, Peak-to-Trough method is applied to data to measure amplitude of the sea bottom reflected wavelets from trace to trace on every CDP gathers group. In input data, all amplitudes were chosen positive polarity. In second step, all determined offset dependent amplitudes are smoothed by moving average operator then, spherical divergence correction is applied to the three CDP gather groups. The decrease in the amplitude of seismic waves due to geometrical conditions is known as spherical divergence and it is discussed in detail in the literature, e.g., O'Doherty and Anstey (1971) and Sheriff (1975). Seismic waves generated at a point source propagate as spherical wavefronts in a homogeneous and isotropic medium. There is a simple relation between increased distance and decreased energy. The energy is decreased as the squared distance increases. Because the amplitude is defined as the square root of the energy, amplitude decreases as the distance increases. Although this simple relation is true for a homogenous, isotropic Earth model, most of the time the velocity change with depth in a layered Earth. In this study, a solution for the divergence problem at an sea/rock media which upper and lower media are homogeneous and isotropic, Newman's equation (1973) was used. All stages given above are preprocess studies. After spherical divergence correction, forward modelling with Trial and Error method is applied to observed data. In forward modelling, input includes central model parameters (lower layer P and S wave velocities and lower layer density), user defined limits (for P, S wave velocities, XI density end Poisson ratio), number of Trial and Error, fit tolerance, number of interfaces, zero offset reflection time, upper media P wave velocity and P wave reflection amplitudes with increasing offset. To estimate density limits of the lower media, Gardner's relation (1974) was used. In determination of Poisson limits, the relation between Vp/Vs and Poisson value for sedimentary rocks is examined. The sea floor of studied area of Saroz Bay is composed of silt and clay. This knowledge gives an important advantage of determination central model parameters. In first step, the model parameters (lower media P and S wave velocities and density) were produced as random numbers. If the physical conditions are satisfied (these are user defined limits), the model parameters are sent to Zoeppritz equations to calculate P wave reflection amplitudes coefficients. The maximum offset was 1450 m. long and the boundary was located at 675 m depth. The model AVO data fall in the precritical region of offset with the maximum angle of incidence of 47° (the critical angle is about 56°). Calculated model AVO data response is normalized to zero offset reflection amplitude. Then, the observed data and model data are compared. If the mathematical constraints satisfied the model parameters are stored. To compare model and observed data, fit tolerance factor is considered. That is, if the produced model parameters can be approached to the observed AVO data with respect to the fit tolerance factor, these parameters are stored. Else, new model parameters are produced and all the stages above are repeated. This trial repeats until the number of Trial and Error. When the number of stored model parameters reached five hundreds, the distribution of solutions for each seismic parameters is calculated. The solutions are normalized. For each distribution function of seismic parameters (lower media P wave velocity, lower media S wave velocity and density), three values are calculated and desired the most likely solutions for each parameter are accepted with 90% confidence intervals among the three values. In the last step, AVO model data (P wave reflection amplitudes coefficients) is calculated with respect to the most likely solutions in the Zoeppritz equations. Then, R(0) zero offset reflection amplitude, error which is the square of between observed and model data differences and Poisson value of lower media are calculated. The results and comments are as follows : * In AVO analysis, considered amplitudes are relative amplitudes. That is, there is no AGC on AVO data. * The basis of AVO analysis is to determine the variation reflection amplitudes with offset. The variation of reflection amplitudes is small in the near offsets but high in the far offsets. Especially, near precritical angle offsets, it takes maximum value. xn * In application, the trend in the AVO data is easily noticeable due to high S/N ratio. As the S/N ratio is increased, the resolving power of the forward method increases and the estimated results approach to the model parameters. Before modelling, to increase the S/N ratio, vertical stack is applied to CDP data. Then, AVO data is smoothed by an operator. * In application, with Trial and Error method, forward modelling is applied to AVO data, and the seismic parameters of Saroz bay sea floor are calculated. Sea floor is composed of unconsolidated silt and clay materials. P wave velocity of these materials is approximately 1575 m/sec., S wave velocity is 850 m/sec., the density is 1.7 g/cm3. The estimated parameters are suitable to the searched values for sedimentary rocks. Adding to this, the calculated R(0) is 0.285 and Poisson value is approximately 0.28. xiu

##### Açıklama

Tez (Yüksek Lisans) -- İstanbul Teknik Üniversitesi, Sosyal Bilimler Enstitüsü, 1997

##### Anahtar kelimeler

AVO analizi,
Deniz tabanı,
AVO anaysis,
Sea floor