Translation, Strain and Rotation: New Approaches to SeismicProcessing and Inversion

点击翻译按钮获取中文摘要

Seismology so far was primarily based on the observation of translational ground motions and the inversion of travel time information. The sensitivity of these measurements (so-called kernels) w.r.t. geophysical parameters is generally distributed along ¿banana¿-shaped volumes from source to receiver. Today we aim at exploiting not only travel times but the complete information contained in the observations (full-waveform inversion). This has become possible as computational hardware allows rapid calculations of wavefields through three-dimensional Earth models. While this is one of the directions that will help improve the quality of tomographic images we propose an additional route. In theory, the motion of a measurement point in a deformable elastic medium has twelve unknowns (three translations, three rotations, and six strains). Based on promising recent theoretical results, we pose a very general question: which combinations of those twelve (point) measurements provide optimal constraints on desired Earth properties. This work is motivated by recent advances in measuring strain and rotations and the availability of such observables (e.g., US Array as direct observation or array-derived). The sensitivities are quantified using principal component analysis. This is complemented by real data studies using US Array observations.