Title:

Seismic and petrophysical studies on seismic wave attenuation

Anelasticity and inhomogeneity in the Earth decreases the energy and modifies the frequency of seismic waves as they travel through the Earth. This phenomenon is known as seismic attenuation. The associated physical process leads to amplitude diminution, waveform distortion and phase delay. The level of attenuation a wave experiences depends on the degree of anelasticity and the scale of inhomgeneity in the rocks it passes through. Therefore, attenuation is sensitive to the presence of fluids, degree of saturation, porosity, fault, pressure, and the mineral content of the rocks. The work presented in this thesis covers attenuation measurements in seismic data; estimation of P and Swave attenuation in recorded well logs; attenuation analysis for pore fluid determination; and attenuation compensation in seismic data. Where applicable, a set of 3D seismic data or well logs recorded in the Gullfaks field, North Sea, Norway, is used to test the methods developed in the thesis. A new method for determining attenuation in reflection seismic data is presented. The inversion process comprises two key stages: computation of centroid frequency for the seismic signal corresponding to the top and base of the layer being investigated, using variable window length and fast Fourier transform; and estimation of the difference in centroid frequency and traveltime for the paired seismic signals. The use of a shape factor in the mathematical model allows several wavelet shapes to be used to represent a real seismic signal. When applied to synthetic data, results show that the method can provide reliable estimates of attenuation using any of the wavelet shapes commonly assumed for a real seismic signal. Tested against two published methods of quality factor (Q) measurement, the new method shows less sensitivity to interference from noise and change of frequency bandwidth. The method is also applied to seismic data recorded in the Gullfaks field. The trace length is divided into four intervals: AB, BC, CD, and DE. The mean attenuation (1/Q_m) calculated in intervals AB, BC, CD, and DE are 0.0196, 0.0573, 0.0389, and 0.0220, respectively. Results of attenuation measurements using the new method and the classical spectral ratio method (Bath 1974, Spencer et al, 1982) are in close agreement, and they show that interval BC and AB have the highest and lowest value of attenuation, respectively. One of the applications of Q measured in seismic records is its usage for attenuation compensation. To compensate for the effects of attenuation in recorded seismograms, I propose a Qcompensation algorithm using a recursive inverse Qfiltering scheme. The time varying inverse Qfilter has a Fourier integral representation in which the directions of the upgoing and downgoing waves are reversed. To overcome the instability problem of conventional inverse Qfilters, wave numbers are replaced with slownesses, and the compensation scheme is applied in a layerbylayer recursive manner. When tested with synthetic and field seismograms, results show that the algorithm is appropriate for correcting energy dissipation and waveform distortion caused by attenuation. In comparison with the original seismograms, the Qcompensated seismograms show higher frequencies and amplitudes, and better resolved images of subsurface reflectors. Compressional and shear wave inverse quality factors (Q_P^(1) and Q_S^(1)) are estimated in the rocks penetrated by well A10 of the Gullfaks field. The results indicate that the Pwave inverse quality factor is generally higher in hydrocarbonsaturated rocks than in brinesaturated rocks, but the Swave inverse quality factor does not show a dependence on fluid content. The range of the ratio of Q_P^(1) to Q_S^(1) measured in gas, water and oilsaturated sands are 0.56 – 0.78, 0.39 – 0.55, and 0.35 – 0.41, respectively. A cross analysis of the ratio of Pwave to Swave inverse quality factors, (Q_P^(1))/(Q_S^(1) ), with the ratio of Pwave to Swave velocities, V_P/V_S , clearly distinguishes gas sand from water sand, and water sand from oil sand. Gas sand is characterised by the highest (Q_P^(1))/(Q_S^(1) ) and the lowest V_P/V_S ; oil sand is characterised by the lowest (Q_P^(1))/(Q_S^(1) ) and the highest V_P/V_S ; and water sand is characterized by the V_P/V_S and (Q_P^(1))/(Q_S^(1) ) values between those of the gas and oil sands. The signatures of the bulk modulus, Lame’s first parameter, and the compressional modulus (a hybrid of bulk and shear modulus) show sensitivities to both the pore fluid and rock mineral matrix. These moduli provided a preliminary identification for rock intervals saturated with different fluids. Finally, the possibility of using attenuation measured in seismic data to monitor saturation in hydrocarbon reservoirs is studied using synthetic timelapse seismograms, and a theoretical rock physics forward modelling approach. The theory of modulusfrequencydispersion is applied to compute a theoretical curve that describes the dynamic effects of saturation on attenuation. The attenuation measured in synthetic timelapse seismograms is input to the theoretical curve to invert the saturation that gave rise to the attenuation. Findings from the study show that attenuation measured in recorded seismograms can be used to monitor reservoir saturation, if a relationship between seismogramderived attenuation and saturation is known. The study also shows that attenuation depends on other material properties of rocks. For the case studied, at a saturation of 0.7, a 10% reduction in porosity caused a 5.9% rise in attenuation, while a 10% reduction in the bulk modulus of the saturating fluids caused an 11% reduction in attenuation.
