Seismic inversion for rock properties (e.g. acoustic impedance) requires a low frequency model, which can be obtained by interpolating well-log data along manually picked seismic horizons. However, such manually picked horizons are subjective to interpreter's understanding of the subsurface geologic structures, and are vertically coarse. In this paper, we will present an automatic method to interpolate well-log data by using seismic painting technique without picking any horizons. The painting horizons show much finer vertical distribution and are data driven. By using this method, we are able to produce a 3D volume from available well-log data without manually picking any horizons. The constructed subsurface model can be filtered into a low frequency model for use in seismic inversion. We apply a previously developed basis pursuit inversion technique on a seismic dataset by using two low frequency models constructed from manually picked horizons and painted horizons. The comparison between two inverted acoustic impedances show significantly better fit to the well-log data from painted horizons, which demonstrates the strong potential of the proposed method for further applications.
展开▼
