Online Journal for E&P Geoscientists

Alsaadan, Sami ^*1; Toksöz, M. Nafi ²; Burns, Daniel ²
(1) Geophysical Technical Services, Saudi Aramco, Dhahran, Saudi Arabia. (2) Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA.

Seismic velocity dispersion and attenuation processes are a direct consequence of the Earth’s inelasticity. Waves propagating in an attenuative medium exhibit amplitude loss and phase change in the seismic wavelet due to the attenuation constant and the phase velocity dependence on frequency. This velocity dependence on frequency (dispersion) is a product of the energy causality when attenuation is present. Higher frequencies travel faster and get attenuated more rapidly than lower frequencies. Attenuation and dispersion significantly reduce the resolution and contribute to the distortion of the seismic image, especially in highly inelastic zones. These processes are dependent on rock properties (e.g., porosity, permeability, pore fluid, fractures, etc.) through which the waves travel. Therefore, the identification, modeling and separation of the components of velocity dispersion can provide key information for predicting the seismic response at different frequencies, and this information can be used in applications for integrated reservoir characterization using seismic data.

A method was developed to estimate the Quality Factor (Q) by identifying and modeling velocity dispersion. This method was applied on a dataset from Gypsy testing site located in Pawnee, Oklahoma. Four subsurface intervals were chosen for this study based on varying rock properties. The first interval was predominantly shale, the second interval was mostly sandstone, and the third interval was made up of shale and sandstone. The fourth interval was a combination of the second and third intervals. Three different data types were used: full wave sonic, bender log, and vertical seismic profile. The estimated central frequencies for the source in each data type were 10 kHz, 1 kHz and 100 Hz, respectively. The modeling was performed using the Discrete Wavenumber method and a logarithmic dispersion relation to calculate a constant Q that best explains the observed velocity dispersion for each of the intervals of interest. The elastic scattering component of the dispersion was negligible. Calculated intrinsic Q of 54, 35, 28 and 30, best explained the field data for the first, second, third and fourth intervals, respectively. Further testing of this method is currently being carried out on another dataset from Saudi Arabia, and the results will be compared with direct Q laboratory measurements from core.