| I first explore viscoelastic features of heavy oil by using ultrasonic velocity measurement data collected over a wide temperature range. I found that bulk modulus is significantly increased compared to conventional oil, which is caused by bulk viscosity. The relation between bulk and shear viscosities is obtained by comparing the viscosity-induced bulk and shear moduli. Furthermore, by using the viscoelastic model of Havriliak and Negami (1967), temperature and frequency dependences of the bulk and shear moduli are predicted.;Density has a large contrast between reservoir and shale and is a desired property for reservoir delineation in the Hangingstone oilfield. A P-P and P-S joint AVO inversion method is developed by extending an Bayesian inversion technique to multicomponent data. The synthetic tests show that the joint method gives us superior results in density estimation compared to P-P only inversion. I applied the developed method to the Hangingstone oilfield to estimate density volume. From the inversion result, a reservoir sand thickness map was successfully obtained.;In addition, a time-lapse AVO inversion method is developed, based on the Bayesian technique, in which all available seismic data are used to obtain initial elastic properties (Vp, V s, and rho) and the changes between two surveys. The method is applicable to incomplete time-lapse multicomponent data sets. After synthetic tests, the method was applied to the Hangingstone oilfield. As a result, estimation of initial elastic properties was improved by using the repeat data. Also, the P-wave velocity change was successfully estimated. The method provides a quantitative description of the reservoir and estimate of the steam distribution.;I establish a rock physics model of poorly-consolidated, heavy-oil saturated sands. For the case of inclusions in a matrix, a generalized singular approximation method is used to obtain the effective properties. Furthermore, the model incorporates the viscoelastic features of heavy oil. Thus, velocity dispersion and attenuation associated with the viscosities can be estimated. Moreover, the elastic property changes caused by steam injection are estimated by changing the fluid properties. The predictions are consistent with different actual measurements, including laboratory, well log, and time-lapse seismic data. |
