Abstract：Technique for reservoir parameter inversion using seismic and log data derives reservoir physical parameters from reciprocal log and cross-borehole seismic data.The technique involves the following major steps:Ascertaining the structural shape of cross-borehole formations. One can derive reflection coefficient section embodying interval velocity interfaces from prestack or poststack seismic data by using Born approximate backscattering linear inversion;and such section is equivalent to prestack or poststack migrated section in effect and is taken as a restraint of cross一borehole structure shape in nonlinear inversion.Obtainment of cross-borehole seismic wave parameters. Parameters such as reflection coefficient,wave impedance and velocity can be estimated respectively by making nonlinear inversion with optimal-fitting Newton method.Formation physical parameter estimation. With the use of neural network technique，borehole physical parameters are derived from log data,and then they are used as the restraints for borehole parameters of nonlinear inversion. After the nonlinear inversion of seismic wave parameters,with the aid of neural network quantitative analysis，one can derive the spatial distributions of formation parameters ineluding porosity,permeability and water saturation degree.Practical software package series of the technique includes versions for I860 computer and advanced workstation. The technique has been used to satisfactorily.process seismic data of 1000km. The technique is characterized by 2. 5-D linear wave equation inversion,and joint use of 2-D nonlinear wave equation inversion and neural network analysis. Geologic horizon resolution is nicely equivalent to that of log data. The technique has been used in some oil fields very successfully.
