Seismic inversion methods are mostly based on gradient-based optimization techniques that minimize the data residual between the observed and the predicted data, which is usually computed from a given estimate of the model. This conventional optimization problem can be reformulated as inversion of skeletonized data, sparse inversion, or multiscale inversion that share the same aim of finding useful solutions by reducing the complexity of either the data space and/or the model space and yet, still use the fundamental governing equations without the need for approximations.

Skeletonized inversion relies on using simplified data sets derived from the original data that retain...

You do not currently have access to this article.