Imaging with multiples using linearized full-wave inversion

Discussion

Note that this technique does not migrate all orders of multiples. It only migrates multiples with a single scattering off the model $m({\mathbf x})$ and other scattering off the sharp boundary in the migration velocity model. Considering that multiples with high amplitude in the data are often generated by sub-surface interfaces of high impedance contrast, this technique can account for most of the significant multiples in the data.

Our method is model-based. One obvious consideration is the accuracy of the migration velocity. A conservative way of applying LFWI would be to put in sharp interfaces that are easy to estimate (e.g. the free-surface, sea-bottom, and top-salt). However, what happens if there is a mis-positioned salt flank in the migration velocity? This may open an avenue for velocity estimation. As an imaging tool, LFWI works as well as other model-based multiple-prediction-subtraction methods.

We recommend applying LFWI for surveys where multiple removal is an issue. An appropriate field study for this method would be a shallow-water dataset with a deep target zone. In this case, each order of multiples overlaps with the previous order, and the conventional multiple-prediction-subtraction techniques might not deliver.

Imaging with multiples using linearized full-wave inversion