Search

ABSTRACT Seismic attenuation and velocity dispersion are potentially able to reveal the rock physical properties of the subsurface. Conventionally, a frequency-independent quality factor ( Q) is measured. This Q is equivalent to the total velocity dispersion in a seismic record and is inadequate for analysing the attenuation mechanism or rock physical properties. Here a new method is proposed to extract the velocity dispersion curves so that more attributes can be obtained from full-waveform multichannel sonic logging data, especially the critical frequency ( fc) if it is within the bandwidth of the data. This method first decomposes the seismic data into a series of frequency components, computes the semblance of each frequency component for different velocity values, cross-correlates the semblance matrices of adjacent frequency components to get the velocity gradients, and finally integrates to obtain a velocity dispersion curve. Results of this method are of satisfactory accuracy and robustness. This method is applied to the data acquired in Mallik 5L-38 gas hydrate research well in Mackenzie Delta, Northwest Territories, Canada. The observed P-wave velocity dispersion compares well with the geological setting. In the gas hydrate zone (about 900 m-1100 m), high concentration of gas hydrate causes very strong velocity dispersion and a distinct fc at about 15 kHz, likely due to strong scattering of centimetre-scale inclusions of gas hydrate; concurrently, water flow in connected cracks in some ranges of this zone adds a large part of velocity dispersion and a dimmer fc at about 9.5 kHz. Immediate underneath the gas hydrate zone, abundant free water in weakly laminated sediments causes quite strong velocity dispersion and an fc at about 6.5 kHz. Velocity dispersion is mild and without an obvious fc in sediments above the gas hydrate zone. [ABSTRACT FROM AUTHOR]

ABSTRACT Quantitative interpretation of time-lapse seismic signatures aims at assisting reservoir engineering and management operations. Time-lapse signatures are thought to be primarily induced by saturation and pressure changes. Core-flooding and reservoir flow simulations indicate that a change of the driving forces during dynamic fluid injection gives rise to a varying saturation scale. This saturation scale is yet another variable controlling the time-lapse seismic signal. In this work, we investigate the saturation scale effect on time-lapse seismic signatures by analysing simple modelling scenarios. We consider three characteristic saturation scales, ranging from few millimetres to metres, which may form during gas injection in an unconsolidated water-saturated reservoir. Using the random patchy saturation model, we compare the corresponding acoustic signatures, i.e., attenuation, reflectivity, and seismic gather associated with each saturation scale. The results show that the millimetre saturation scale produces minimum attenuation and the same seismic signatures with those obtained from the elastic modelling. The centimetre saturation scale produces maximum attenuation, whereas the metre saturation scale causes highest velocity dispersion. The analyses of the time shift and amplitude change indicate that ignoring a time-dependent saturation scale can result in biased estimation/discrimination of the saturation and the fluid pressure. In particular, the 4D signal can be strongly affected by the saturation-scale change when the reservoir gas saturation is low and the effective pressure is high. In the presence of an increasing (decreasing) saturation scale during injection, interpreting an observed time shift and amplitude change using the Gassmann model will lead to underestimation (overestimation) of the change in gas saturation and fluid pressure. We show that including the effects of capillarity and residual saturation into the rock physics modelling can potentially reduce the interpretation uncertainty due to the saturation-scale change. [ABSTRACT FROM AUTHOR]

We have analysed the local earthquakes (2.0 ≤ ML ≤ 5.5) occurred in northeast (NE) India recorded by a temporary seismic network of 10 stations operated by National Geophysical Research Institute (NGRI), Hyderabad to evaluate the relative contributions of scattering loss ($Q_{{\rm sc}}^{ - 1}$) and intrinsic absorption ($Q_{\rm i}^{ - 1}$) to total attenuation ($Q_{\rm t}^{ - 1}$) using the multiple lapse time window analysis assuming multiple isotropic scattering in a medium of uniformly distributed scatterers. The results show that $Q_{\rm i}^{ - 1}$ is greater than $Q_{{\rm sc}}^{ - 1}$ at high frequencies (f > 3 Hz), while the opposite is observed at low frequencies (f < 3 Hz). The observed frequency dependence of $Q_{{\rm sc}}^{ - 1}$ corresponds to the scale length of lithospheric heterogeneity beneath NE India, at least comparable with the wavelength for the lowest frequencies analysed, of about 1 km. The observed $Q_{{\small c}}^{ - 1}$ for the study region obtained with single scattering theory is close to $Q_{\rm i}^{ - 1}$ at high frequencies, in agreement with theoretical prediction for an idealized case of uniform distribution of scatterers. However, a discrepancy exists between the two at low frequencies, which can be explained by a depth-dependent velocity and attenuation structure. High value of $Q_{\rm t}^{ - 1}$ is correlated with the geology and tectonic settings of the region characterized by Himalayan and Burman collision zones with a strong lateral heterogeneity. The $Q_{\rm i}^{ - 1}$ estimates obtained in this study can be used to infer the average temperature of the lower crust with an upper limit estimate of ∼800 °C assuming a lower crustal gabbroic lithology. [ABSTRACT FROM AUTHOR]