Through time, deterministic methods have been used for the prediction of reservoir properties. A relationship that involves the observed seismic data and the reservoir parameter of interest is determined. This transformation is used to convert seismic data to the desired reservoir property.
In the last years, non-linear multivariate determinant analysis using neural networks between multiple seismic attributes and the measured gamma-ray values has been used to produce inversion volumes of the desired log type. This kind of approach has been very useful in gas exploration and enhanced recovery.
This work discusses the computation behind the reservoir properties.
The mathematical relationship between pairs of variables derived from well logs and then transforming the 3D seismic attributes volumes into the desired reservoir parameters (Chopra, Alexeev, & Pruden, 2004).
for full SEG text: