next up previous print clean
Next: Least-Squares Joint Imaging of Up: Brown: LSJIMP field data Previous: Brown: LSJIMP field data

Introduction

Multiple reflections are generally treated as noise in the reflection seismic experiment, and strategies for their removal occupy a prominent place in the exploration seismic literature. Tantalizingly, though, multiples often penetrate deeply into the earth, and thus have the potential to aid the imaging of the prospect zone. Algorithms to structurally image multiples have been in the exploration geophysics literature for over a decade Berkhout and Verschuur (1994); Guitton (2002); Reiter et al. (1991); Shan (2003); Yu and Schuster (2001), though none successfully addresses perhaps the most important question: How can any usable information from the multiples be rigorously extracted and exploited? In a previous work Brown (2002), I presented a least-squares joint imaging technique which tentatively answers this question.

In two companion papers in this report Brown (2003a,b), I derived operators which, to reasonable accuracy, model the physical (kinematic and amplitude) connection between pegleg multiples and their corresponding primary events. These operators transform a pegleg multiple into a ``copy'' of its primary. My regularized least-squares joint imaging scheme, ``LSJIMP'', (Least-squares Joint Imaging of Multiples and Primaries) then exploits this redundancy to separate the pegleg multiples from the data, and at the same time, ``spreads'' additional information provided by the multiples into the primary image.

I apply LSJIMP to the Mississippi Canyon 2-D multiples test data, and it demonstrates good noise removal and signal preservation characteristics. LSJIMP's inherent efficiency, combined with the use of Message Passing Interface (MPI) parallelization ensure that its computational performance compares favorably with other advanced multiple suppression techniques.


next up previous print clean
Next: Least-Squares Joint Imaging of Up: Brown: LSJIMP field data Previous: Brown: LSJIMP field data
Stanford Exploration Project
7/8/2003