Wide-Angle AVO Waveform Inversion with WKBJ Modeling

We propose a new target-oriented inversion workflow aimed at extending AVO-type inversions to long-offset seismic data. Such an ability improves reservoir parameter estimation, allowing sharper discrimination of velocity and density changes across an interface in comparison with estimations done with AVO inversion based on plane-wave reflection coefficients. The forward modeling used during the inversion is the WKBJ seismogram algorithm, a fast and robust way to compute seismograms from pre-critical partial reflections, through transitional waveforms, which include the interference with head waves, to totalreflection signals. The inversion fits a target reflection in raw shot or CMP gathers by direct comparison of either the waveforms of the observed data and<br>modeled WKBJ seismograms or of their respective envelopes. In addition, to avoid local minima the workflow has been divided into two stages: first, a global random search of the parameter space provides the initial model, which is then fed into the second<br>stage, a local gradient-based optimization. Inversions of a synthetic model and a raw shot gather from a 2D point-receiver survey demonstrate that the inclusion of wider-angle, near-critical and post-critical data in the proposed inversion improves the resolution of both velocity and density contrasts.