| Three seismic attributes commonly used for pore-fluid and lithology prediction are the fluid factor, Poisson impedance (PI) and lambdap. While the sensitivity of one attribute versus the other are often claimed to be better than the other, I find that the three attributes are not independent but can be traced back to the fluid factor. As a means of testing the sensitivity of the three attributes, rock-property attributes for different lithologies and pore fluids and the corresponding rock-property statistics were derived from 183 suites of well-log data and used to calibrate the borehole data to similar attributes derived from the seismic data. Normalization from seismic amplitude to well-log data makes it possible to quantitatively predict the pore-fluid based on the well-log dataset.;Numerous inversion schemes, which require additional information such as low-frequency trends, have been developed to quantitatively measure formation layer properties derived from seismic. However, quantitative reflectivity interpretations have been hindered by the inability to convert seismic amplitude into meaningful reflection coefficients. I have developed a method of accurately predicting the normal-incident reflection coefficients (NI) based on pore-fluid and lithology transforms derived from well-log curves in the area. The interpretation is based on a comparison of reflection coefficients computed for the prospect and its down-dip equivalent brine-saturated reservoir. The down-dip wet area for the seismic data is then converted to its equivalent NI for gas and oil saturation. Histograms and CPDFs of NIP for wet, oil, and gas in the down-dip area are used to interpret the pore fluid at the prospect from its NIP. Bayes's decision rule is used for the final classification of saturation for the reservoir. The methodology accounts for amplitude variations caused by thickness changes in the seismic survey. Seismic and well-log data in South Marsh Island, Gulf of Mexico are tested by this method.;An offset-dependent NMO correction is proposed for processing seismic data across faults. During NMO corrections, instead of using the velocity at the CMP location, the NMO velocity for each offset trace is derived from an interpolation of the velocity functions at the source and receiver locations. This approach flattens events when the CMP raypaths pass through a fault block with significantly different velocities for hanging wall and foot wall. |
