Sparse Seismic Deconvolution by Method of Orthogonal Matching Pursuit

Abstract

The method of orthogonal matching pursuit is used to solve the sparse deterministic deconvolution problem for poststack seismic data. A description of the theory is given and algorithmic implementation is discussed. This method is then applied to synthetic and real data. The orthogonal matching pursuit method, like other sparse spike techniques, produces high apparent resolution (spiky appearance) in the output. What value this and other such sparse spike estimators might bring to stratigraphic interpretation, if any, is still under assessment.