Posted by Werter Silva

Seismic imaging is a powerful tool for visualizing and characterizing subsurface geologic structures. It has been shown that this process can be formulated as an inverse problem, where the recorded seismic data are linearly related to model perturbations using the Born approximation. However, the seismic data require multiple attenuation prior to imaging due to the single scatter assumption embedded in common imaging technology. Conversely, nonlinear imaging uses solutions to an inverse problem that aims to recover model parameters from entire data without multiple attenuation. By reparametrizing the wave equation, this technique can solve for parameter contrasts that act as proxies for the seismic image. This allows multiples to contribute to imaging, resulting in cheaper and higher-quality images with a reduced turnaround time, given that multiples do not require prior attenuation.

The image above compares seismic images obtained by nonlinear inversion with RTM and least-squares RTM images. The input data for all of them is a single shot gather for a source positioned in the middle of the horizontal axis containing primaries, surface, and internal multiples. The nonlinear inverted image (d) does a good job of recovering the two reflectors and increasing their resolution compared to the RTM (b); it also shows a better balance in amplitude since both synthetic reflectors have the same strength. The ghost “derivative” effect, seen more clearly in the LS-RTM image (c), is accounted for, producing a symmetric pulse. It works better for near offsets due to the radiation pattern of the adopted parametrization. With more shots in a complex model, the staking effect is expected to reduce the relative strength of the data residual. This promising imaging technology still needs to be tested on more complex models.