In acoustic exploration and monitoring imaging plays a critical role in uncovering structure and minute changes therein. It is, however, often hampered by unfulfilled assumptions. One such assumption is that estimated incident and reflected wavefields at a reflector travel in opposite directions with respect to the reflector surface-normal vector. In Reverse Time Migration (RTM) this is often not the case due to the common use of scattering modelling operators to estimate incident and reflected wavefields at an interface. This results in artefacts in the output RTM image. A partial remedy to these artefacts is to directionally decompose wavefields during the RTM imaging step. This allows for the decomposition of the incident wavefield at a horizontal interface into a down-going wavefield, which in conjunction with the upgoing reflected wavefield can be used to estimate the reflectivity at the interface. Directional wavefield decomposition classically decomposesmulticomponent wavefields recorded along a horizontal surface into up- and down-going wavefields. However, not all techniques implicitly use wavefields recorded along flat surfaces. RTM, for example, commonly works on snapshots of a wavefield, hence the classical decomposition techniques are hardly applicable. Other techniques have been developed to solve this problem, but still only decompose into up- and down-going wavefields as it is assumed that the media of interest only vary in the vertical direction. None allow for an elegant decomposition of a wavefield according to all possible travel directions. In this thesis we develop a snapshot acoustic directional wavefield decomposition technique with an emphasis on the fact that the method works on snapshots of wavefields in timeand that the directionwith respect towhich the decomposition occurs is arbitrary, and not simply the vertical direction as in up-down decomposition. We demonstrate how to directionally decompose an acoustic wavefield according to the directions of propagation of its constituent plane waves by showing how to separate a wavefield into its constituent plane waves. This allows for approximate wavefield decomposition in arbitrary media, even normal to interfaces. This is an obvious boon for imaging complex structures, as imaging should occur normal to surfaces and not only in the vertical direction.
|Qualification||Doctor of Philosophy|
|Award date||18 May 2020|
|Publication status||Published - 18 May 2020|
- Wavefield Decomposition