IRANIAN JOURNAL OF GEOPHYSICS
Year:2009 | Volume:3 | Issue:1|Papers Count:7

In this study, an attempt is made to predict effective porosity in one of the oil fields in the Persian Gulf by designing a probablistic neural network (PNN) and simultanusely making use of seismic attributes and effective porosity logs in the reservoir window. This was done by deriving a multiattribute transformation between an optimum subset of seismic attributes and the effective porosity logs.The geophysical data used in this study consist of 3D seismic pre-stack time migrated (PSTM) data with 12.5*12.5 m grid size and a 4 ms sampling rate. The length of the seismic traces are two seconds. Well logs of five vertical wells in the study area, including Sonic (DT), Density (RHOB), Effective Porosity (PHIE) and Seismic Well Velocity Surveys (Check Shots), were used. The reservoir layer is a Mishrif member of the Sarvak formation with Cretaceous age, which is common in oil reservoirs in the Persian Gulf. The top of the Mishrif is adjusted with the Middle Turonian Unconformity and covered with shaley Laffan formation. The Mishrif Reservoir in study area contains two reservoir zones. The lower zone with higher clay content is separate from the upper zone. The upper zone consists of clean limesone with better reservoir properties. Seismic traces close to the well locations were used to generate seismic attributes. Effective porosity logs at the reservoir area were the target logs in this study.The designed neural network consists of one input layer, one hidden layer with four processing units (neuron), and one output layer with one neuron. In order to prepare training samples for the neural network, PHIE logs were converted to time domain using a time-depth relationship calculated from the DT logs and check shot curves for each well location. Subsequently, these logs were filtered (using a Hanning filter with 4 ms length) and resampled with seismic sampling rate (4 ms). Finally, a set of seismic attributes, including sixteen sample-based seismic attributes, were generated using HRS software. Training samples in this study consisted of 57 samples (selected seismic attributes and their related effective porosity from PHIE logs in the time domain). For training the network, the samples were divided into three data sets: the training samples, cross validation samples and testing samples. The training data were used for adjusting the weights of the network; the cross validation data were used to prevent overtraining theneural network; and the testing data were used to ensure generalizabillity of the network output.A forward stepwise regression process was used to determine an optimum subset of attributes for use in the training of the neural networks. The optimum subset of attributes in this study consists of the Dominant Frequency, Amplitude Weighted Frequency, Integrated Absolute Amplitude and Filter 45-60 Hz.After the network was trained using training and cross validation data sets, it was used to predict the testing data. The results show a good correlation between real and predicted data, with 92% correlation. Finally, in order to attain a better generalization of the network, testing data sets were inserted to trained data and the network was trained again. This network was then used to predict effective porosity in well locations which increased the correlation coefficient to 95%. This study shows the ability of the PNN networks to predict effective porosity even with a paucity of training examplares.

Seismic waves lose energy by traveling through the earth. Attenuation refers to the loss of energy which is caused by parameters other than geometrical spreading, and depends on the characteristics of the transmitting medium. Generally, attenuation is determined by quality factor (Q) which is a dimensionless parameter and has a reverse relation with attenuation coefficient. Experiments show that seismic wave quality factor depends on the fluid content of formation and its elastic properties. Hence, quality factor (Q) (or attenuation) is one of the most important attributes in seismic exploration used as a direct hydrocarbon indicator. Attenuation coefficient is usually calculated in frequency domain based on power spectrum and statistical methods. Because of the Fourier transform's limitations in analyzing non-stationary signals, this paper proposes an approach using three different harmonic analysis methods and two different attributes, namely the centroid of scale and the centroid of frequency, to determine the seismic quality factor (Q). The three methods consist of the continuous wavelet transform (Mallat, 1999) with modified Morlet wavelet, the smoothed pseudo Wigner-Ville distribution, and the reassigned form of the smoothed pseudo Wigner-Ville distribution (Auger and Flandrin, 1995). Because the Wigner-Ville distributions are energy distribution type, they were used in this study to determine the seismic quality factor. Since both the traditional Wigner-Ville distribution and pseudo Wigner-Ville distribution suffer from cross terms, the smoothed and reassigned forms of them were implemented to overcome the cross terms. Smoothing extends the auto terms in Wigner-Ville distribution which is not desirable for the purposes of this study. Therefore, by using the reassigned method we solved the problem of auto terms extension in the smoothed pseudo Wigner-Ville distribution. According to the results of this study, the reassigned smoothed pseudo Wigner-Ville distribution indicates fewer cross terms than traditional Wigner-Ville and provides better resolution than smoothed pseudo Wigner-Ville distributions.We can verify that for a spike, as an ideal seismic source wavelet, the centroid of scale is inversely proportional to the quality factor (Equation (1)), while the centroid of frequency has a direct relation with the quality factor:sc(t)=mt/QwhereScis the centroid of scale, is time, is the modulation frequency, and is the quality factor. For a band-limited seismic wavelet, the estimated quality factor using Equation (1) differs from the true value. In such cases, there will be a relative reverse relation between the centroid of scale and quality factor as well as relative direct relation between the centroid of frequency and quality factor. Therefore, it is preferred to use the centroid of scale or centroid of frequency as a qualitative attribute for Q-factor determination. tmQ The efficiency of the introduced methods was investigated on both synthetic and real seismic data. The results showed that the quality factor estimated by the reassigned smoothed pseudo Wigner-Ville distribution method has better resolution than that of the other two transforms. Furthermore, the frequency-based results for the estimated quality factor show the low frequency shadow properties beneath the true position of the anomaly while the same results based on amplitude show the anomaly at its true position. Moreover, the results indicate that the existence of noise in the data do not affect the efficiency of the methods.

Acoustic waves propagated in gas and lossy media are less attenuated and more reliable than elastic waves. Accordingly, the detection and measurement of acoustic waves are more precise than for elastic waves. Consequently, reservoir characterization using processing techniques and seismic data inversion are based on propagations of acoustic waves. The wave field in acoustic media is described by a scalar quantity rather than by a vector. The tau-p transformations in the near field for the acoustic and elastic wave equations are similar, as they both yield the eikonal equation in isotropic media. The reflection and transmission behaviors of waves, however, differ considerably in each of the two media. In acoustic media, all P-wave energy is conserved and as a result it can be used for near offset tau-p modeling.Fryer (1980) developed a reflection method for modeling the VSP data in tau-p domain. The main idea of using tau-p domain is to investigate the amplitude variations as a function of near offset, changes of phase, estimation of attenuation factor (1/Q) and separation of primary multiples. Laboratory measurements and well log data have shown that the Q factor depends on the type of media and also the percentage of saturation. Therefore, Q is a very strong factor for characterization of reservoir gas zones. Due to its sensitivity, this factor is used in our two different modeling programs.In this study, the VSP modeling was used in the t-x and the tau-p domains. One of the most important parameters in this modeling is the velocity model. To generate the velocity model and synthetic seismograms, a computer program was developed in t-x and tau-p domains. Then, for two exploration wells, the VSP models were compared using real data. Based on the above algorithm, a software package was developed using a finite difference method in the tau-p domain. The slowness-time reflectivity method was used to calculate tau-p synthetic seismograms with the inclusion of the attenuation factor (1/Q) in lossy media. For large acoustic impedance contrasts, the attenuation factor occurs as an amplitude decay and phase rotation for some range of high frequencies. First, the upcoming and downgoing VSP synthetic data for side locations along each well were modeled and compared with the VSP real data. Second, the normal incident seismic sections based on well logs were compared with the tau-p sections derived from the acoustic and viscoacoustic media. The VSP data and seismic modeling techniques were used for the detection of gas zones in two wells in one of the south Iranian reservoir. The models of traveltime and amplitude changes of VSP data in the tau-p and in the t-x domains proved to be effective techniques for detection of gas zones from the VSP seismic data. Using this technique, gas zones in the reservoir can be very reliably detected using acoustic waves. The aforementioned procedure were applied in verification of different media.A comparison of the VSP data generated in the tau-p and the t-x procedures can yield valuable results. Results show that modeling in the tau-p domain using a localized slant stack is faster and more reliable than are conventional methods. Additionally, wave energy characteristics and amplitude changes of seismic waves in two different acoustic and viscoacoustic media were investigated using a 2D acoustic wave algorithm in lossy media. By using acoustic, viscoacoustic and anisotropic models in the tau-p domain and comparing them with normal reflections and VSP data, one can detect saturated gas zones. The synthetic VSP in the tau-p domain definitely helped to verify changes in amplitudes and phases in two VSP well data sets investigated here. Using this technique, it was found that reservoir gas zone can be reliably detected by acoustic waves. Furthermore, it was established that these waves can be used for better comparison with real VSP data in the reservoir gas zones.

Pore volume compressibility, which is the reverse of bulk modulus, is one of the most important and effective parameters of mechanical, seismic and reservoir properties of hydrocarbon reservoirs. The investigation of elastic modulus is essential for using geophysical data in reservoir production and EOR. In order to determine the pore compressibility of Asmari carbonate reservoir rocks, a total of 90 samples from different types of carbonate rocks were selected from oil reservoir for this study. Petrographical analyses were conducted to determine the effect of textures; type and value of porosity on pore compressibility, and then the effect of net confining pressure and pore compressibility on the Archie cementation coefficient was studied. To achieve this purpose, after geological examinations and grouping the samples based on type of textures and pores, the relationships between cementation coefficient and pore compressibility were calculated.Pore compressibility experiments as well as the determination of the cementation factor were performed using the “Overburden FRF Rig” apparatus. Hydrostatic pressure up to 10000 psi may be applied to samples via this apparatus. The sample that is fully reservoir brine saturated is located in a rubber sleeve of hydrostatic core holder and surrounded by two rings. After purging the system, the confining pressure was increased from 435 psi up to 4000 psi. The increase of pressure on rocks causes pore compaction and expels the fluid from the sample. The volume of fluid expelled from the samples shows the volume of pore volume reduction due to the increase of pressure. Pore compressibility was determined by derivation of each pressure point by plotting the logarithm of variation of pore volume reduction vs. effective confining pressure. In order to study pore compressibility under hydrostatic pressure, five steps of increasing the net pressure, which were 435, 1000, 2000, 3000 and 4000 psi, were selected.Pore compressibility is a function of porosity and it increases as porosity decreases. The power function, expressed as Cp=a(P)b, usually shows the best fit for the compressibility and net pressure relationship.The results of this study show that for almost all of the selected samples, the cementation factor increases when the pressure is increased. The compaction of rocks causes changes in the structure of the pores and grain shapes, and reduces the pore volumes. In a few of the samples, increasing the pressure caused a reduction in the cementation factor. The main reason for this reduction could be due to damage to the structure of the rocks.Two main differences were observed between the changes in compressibility and cementation factor: (1) by increasing compressibility, the reduction of the cementation factor is accelerated in samples with a higher cementation factor and (2) changes of pore compressibility are higher for samples with a lower cementation factor.Isolated porosity and separated vug porosity are closed by a slight increase of pressure in comparison with connected porosity, and therefore reduction in cementation factor is correlated with low pore compressibility. An increase of dolomitization in the samples from packestone to dolopackestone and dolostone causes better connections among pores and the effect of pore compressibility is more pronounced. A summary of results is as follows: 1) The slope of curve increased in a cross plot of the measures of cementation coefficient vs. compressibility due to a change in texture from packestone to dolopackstone and dolostone.2) Increasing the net confining pressure increases cementation factor.3) Variation of pore compressibility is higher in the samples with lower a cementation factor.4) The cementation factor is reduced by an increase of pore compressibility, and reduction of cementation factor occurred faster in the samples with a higher cementation factor.5) In the investigated formation, the value of pore compressibility is dependent on the texture, type and value of porosity.6) Pore compressibility is higher in sandy dolostone with porosity measuring less than 15 percent compared to other carbonate textures, and it decreases with change in texture from sandy dolostone to mudstone.A power function in the form of Cp=a (P)b usually shows the best fit for compressibility and the net confining pressure relationship.

Teleseismic body wave tomography beneath a profile of portable seismic stations using the ACH method (named after authors Aki, Christoffersson, and Husebye) is generally based on relative residual data from teleseismic earthquakes. The relative residuals are inverted to retrieve the two dimensional structure of the velocity perturbation relative to a spherical reference Earth model (for example, IASP91) in the structure of interest beneath the profile. This method tries to minimize the influence of extraneous factors, such as errors in earthquake location or origin time and ray paths from the source to the base of the target volume, by subtracting the mean of the arrival-time residuals for each event, since only the velocity deviations in the target model are investigated. The data are then corrected for crustal travel-time variations a perior inversion. Because travel time perturbation reflects the velocity perturbation integrated along the ray path, some perturbations in the target model may be caused by deeper structures in the upper mantle. This paper intends to study whether the de-meaning process used in the ACH method can remove the effects of deeper mantle anomalies (especially those located directly underneath the target region) or deeper heterogeneities that may leak into the velocity structure of the region of interest. Therefore, considering some different hypothetical velocity structures, including positive and negative anomalies (with relatively high and low velocities, respectively), this study targets an area approximately 1% immediately underneath the base of the model at a depth of 460 to 660 km in an attempt to determine how the velocity structure of the upper mantle beneath a profile would be affected by the presence of possible anomalies in greater depths. Hypothetical tests were applied using teleseismic data recorded in a profile across the Zagros collision zone. The Zagros seismic experiment comprised 66 short-period, medium- and broad-band stations deployed along a NE-SW transect from Bushehr to Posht-e-badam in the southwestern part of Central Iran between November 2000 to April 2001. The profile is believed to be almost perpendicular to the main tectonic units of the Zagros collision zone. For the target model, a simplified P-wave structure to a depth of 460 km based on the tectonic observations and previous tomographic results consisting of two relatively high and low velocity anomalies of approximately±3% at depths of 120 to 300 km, respectively, beneath the Zagros zone and Central Iran were embedded within the reference Earth model. These two anomalies were separated by a sharp sub-vertical transition. Using the target model structure and postulating different anomalies underneath the base of the model, the relative residuals were inverted. The results indicate that these hypothetical heterogeneities in the mantle below the base of the model leads to some effects in the velocity structure of depths lower than 300 km, which have lower resolution. These effects could be attributable to insufficient resolution of the target model at these depths due to a low number of criss-crossing rays. Moreover, there are some deviations in depths of 120 km up to Moho. However, all models retrieve the major features (including transitions and major blocks) available in the hypothetical model, albeit with underestimated amplitude due to the regularization parameters and model parameterization.

Iran is located in a very complex tectonic area, where continental shortening takes place due to the collision of the Arabian and Eurasian plates. This Arabia-Eurasia Convergence occasionally causes a distractive earthquake such as the Silakhor Earthquake to occur in Iran. The March 31 2006 Silakhor Earthquake, with a magnitude of Mn=6.0, occurred on Friday at 4: 47: 03 local time near the village of Chalan-Cholan in Lorestan province. It was preceded by two large foreshocks with magnitudes of Mn=4.7, 5.2 and followed by two relatively large aftershocks of Mn=4.9 and 5.3. The Silakhor plain was seriously affected in earthquake: about 70 people were killed and more than 2000 were injured (Mirzaei Alavijeh et al.1385). Mahdavifar and Tajik (1385) reported a macroseismic intensity of I0=VIII on the MSK98 scale for the Silakhor Earthquake.More than 30 foreshocks and many aftershocks were recorded in the Silakhor Earthquake. Such extensive foreshock-mainshock-aftershock sequences for an earthquake of moderate. magnitude (Mn=6.0) is unusual The earthquake sequence occurred along the Main Recent Fault (MRF) in the northern Zagros. The right-lateral strike-slip displacement along the MRF fault is about 3±2 millimeter per year (Vernant et al.2004). All seismic stations which recorded Silakhor earthquake are located at a distance of at least 100 kilometers from the epicenter. No attempts were made to record micro events by deploying a temporary seismic network in the source region. To analyze this earthquake, an approach was made to relocate the recorded sequence by gathering all available data and using a proper velocity model. The region in this study enclosed between 48.4o to 49.2o east longitudes, and 33.3o to 34o north latitudes.The Silakhor earthquake sequence was recorded in many permanent seismic stations.These stations are operated by the University of Tehran's Institute of Geophysics (IGUT), the International Institute of Earthquake Engineering and Seismology (IIEES), and the Karkeh Dam. All picked phases data were compiled from the bulletins of organizations listed above. Additionally, in order to plot a PGA contour map of the source region, the acceleration data recorded in stations which are run by the Building and Housing Research Center (BHRC) were used.For this research, recorded events were relocated by using different location programs and a proper velocity model. The results show that applying the Hypoinverse program (Klein, 1984) gives less errors for hypocenter parameters in comparison with other location programs, such as Hypo71 (Lee and Lahr, 1975) and Hypocenter (Lienert et al. 1986). The focal mechanism of the main shock was determined by using the polarity data of the first arrival waves in the seismic stations. The mechanism for the mainshock was obtained as Strike/Dip/Rake=310o, 46º, 171o. The obtained focal mechanism shows that the activated fault segment in this earthquake has a right lateral mechanism with a dip toward the north-east. The focal mechanism of mainshock and iso-acceleration curves show that the mainshock rupture in the Silakhor Earthquake was a unilateral rupture and that it initiated near the southeastern end of the rupture zone and propagated toward the northwest. The distribution of relocated aftershocks shows that in the area of this earthquake two fault segments were active. The first segment, which includes the main shock, is located between Dorud and Silakhor. The second segment is located between Silakhor and Borujerd. It seems that there is a gap of about 10 km length between them.

The Mediterranean climate is characterized by dry hot summers and rainy winters. In the summer, the Northern Hemisphere high pressure belt migrates northwards, dominating the Mediterranean area. During the winter, the high-pressure belt drifts back towards the equator, and the region becomes more dominated by rain-bearing cyclones. Apart from the seasonal changes in the frequency and intensity of cyclones, the track of the cyclones varies with the season, tending toward the south-east in the summer and to the east in the winter. The main cyclogenesis centers in the Mediterranean are the Gulf of Genova, the North Red Sea, Cyprus, West Mediterranean, East Mediterranean, Southern Italy, and the Iberian Peninsula.In the present study, the effects of the annual frequency of cyclones generated in different Mediterranean centers and the annual mean sea-level pressure of the centers on annual precipitation of Iran are investigated using data for the period of 1960 to 2002. The geographical distribution of correlation coefficients between the precipitation in Iran and the frequency of cyclones in the Mediterranean cyclogenesis centers indicates that the annual cyclone frequency in the Mediterranean area significantly affects the annual precipitation of almost all regions of Iran, except those in the southeastern, eastern and central parts of the country. The two cyclogenesis centers that affect the precipitation of larger areas of Iran are the East Mediterranean and the North Red Sea. The regions of Iran under the influence of these two centers are located in the north, north-west and south. While precipitation in the south-east and central regions of Iran has no significant correlation with the sea surface pressure at any of the Mediterranean centers, those in the west and north-west are significantly correlated with the sea-level pressure of almost all of the Mediterranean centers.Two sets of multi-variable linear regression models were developed to regress the annual precipitation from stations throughout Iran to 1) the number of cyclones generated in different Mediterranean centers (cyclone frequency models) and 2) the mean sea-level pressure at representative stations of the Mediterranean centers (sea level pressure models). Correlation coefficients between the measured annual precipitation and that predicted using the cyclone frequency models for the stations in the west, east and center of Iran are significant, and the standard error of estimation is smaller in the west than in the east. The correlation coefficients between the observed annual precipitation and that predicted by the mean sea-level pressure models are significant for stations located in the center, northeast, east and southeast of Iran. The precipitation calculated by the sea-level pressure models is closer to observations compared to that predicted by the cyclone frequency models. During winters with negative precipitation anomaly (drought) in Iran, the subtropical high pressure belt (the Azores high pressure) dominates the Mediterranean and the mean sea-level pressure is anomalously high. This situation decreases the frequency of cyclones in almost all of the Mediterranean centers.The correlation coefficient between the cyclone frequency and sea-level pressure is significant only in the West Mediterranean and Gulf of Genova centers. It seems there are two specific features that distinguish these two from other Mediterranean centers: a) the migrating cyclones generated in other centers do not pass through these centers and b) the geographical distribution of the correlation coefficients between precipitation in Iran and cyclone frequencies and sea-level pressure of these two centers are very similar.