Fast sweeping algorithm for accurate solution of the TTI eikonal equation using factorization

Umair bin Waheed, Tariq Ali Alkhalifah

Research output: Contribution to journalArticlepeer-review

43 Scopus citations

Abstract

Traveltime computation is essential for many seismic data processing applications and velocity analysis tools. High-resolution seismic imaging requires eikonal solvers to account for anisotropy whenever it significantly affects the seismic wave kinematics. Moreover, computation of auxiliary quantities, such as amplitude and take-off angle, rely on highly accurate traveltime solutions. However, the finite-difference based eikonal solution for a point-source initial condition has an upwind source-singularity at the source position, since the wavefront curvature is large near the source point. Therefore, all finite-difference solvers, even the high-order ones, show inaccuracies since the errors due to source-singularity spread from the source point to the whole computational domain. We address the source-singularity problem for tilted transversely isotropic (TTI) eikonal solvers using factorization. We solve a sequence of factored tilted elliptically anisotropic (TEA) eikonal equations iteratively, each time by updating the right hand side function. At each iteration, we factor the unknown TEA traveltime into two factors. One of the factors is specified analytically, such that the other factor is smooth in the source neighborhood. Therefore, through the iterative procedure we obtain accurate solution to the TTI eikonal equation. Numerical tests show significant improvement in accuracy due to factorization. The idea can be easily extended to compute accurate traveltimes for models with lower anisotropic symmetries, such as orthorhombic, monoclinic or even triclinic media.
Original languageEnglish (US)
Pages (from-to)WB1-WB8
Number of pages1
JournalGEOPHYSICS
Volume82
Issue number6
DOIs
StatePublished - Aug 21 2017

Fingerprint

Dive into the research topics of 'Fast sweeping algorithm for accurate solution of the TTI eikonal equation using factorization'. Together they form a unique fingerprint.

Cite this