
Full text loading...
Low-frequency information is very critical in quantitatively predicting porosity, fluid content, and other reservoir properties, however, seismic data with bandwidth constraints lack low-frequency attributes. Compensation of low frequency is very important for obtaining absolute values of rock properties. Conventional low-frequency impedance (LFI) models are usually built by laterally interpolating and extrapolating impedance logs between well sites. This interpolation driven by the distance between wells and guided by interpreted horizons, often leads to artifacts and generation of non-geologic solutions. Velocity data partially provide the missing information in the lowest frequency range. In this paper, we apply the technique of multi-attribute linear regression to find the relationship between seismic attributes (such as interval velocity, relative impedance) and rock properties (acoustic impedance) from well logs. The relation is then used to predict low-frequency P-impedance away from well sites. Tests of the 2-D synthetic Marmousi2 dataset and 3-D field datasets show that this technique is able to obtain a reliable, low frequency, P-impedance model in areas with complex geological structure.