Abstract:Log-constrained polynomial fitting extrapolation inversion of post-stacked seismic data makes model responses coincide with observed data under feasible priori constraint.The inversion results have desirable resolution improvement and offer reliable seismic parameters of finite geological meaning.This method is very significant to thin interbed correlation,reservoir description and reservoir prediction.This article details the polynomial fitting extrapolation inversion using both least-square algorithm and equal(or unequal)weight log-constraint.The constraint error deviations between different layers are considered here,so that inversion resuits are consistent with real geological structures(such as occurrence and structure of formation).As a result,in the area where there are logging data,we can produce reflection coefficient and wave-impedance sections which nicely correlate to logging data and real stratigraphic structure.