| 摘要 |
Method simulating local prestack depth seismic migrated images from target models, without using either real or synthetic recorded data. The input is a background model and some surveys, with the possibility of defining some acquisition surfaces to describe any acquisition geometry of potential surveys. In addition, detailed target models are given, generated from different type of input, such as parameter grids, interpreted time- or depth-horizons with attributes reservoir models, and other models. In the most efficient application of the invention, a point in the background model is chosen by the user and will act as a node for Green's functions calculation between the surveys/acquisition surfaces and that point. Green's functions can be calculated in many ways (classic ray tracing, Wavefront Construction, and Eikonal solvers are possible methods), the mandatory information being slowness vectors to form a sum vector called the scattering wavenumber. According to some survey choices, the scattering wavenumbers are extracted, with various sub-selections, sorting, re-ordering, and used to create filters in the scattering wavenumber domain. Amplitudes, pulses, scattering pattern, Fresnel-zone effects, and other effects can be included in the filters. The latter are then applied to each target model to give a simulated prestack local image in depth after some transformation, like Fast Fourier Transform, from the wavenumber domain to the space domain. For better accuracy, local images for neighboring Green's functions nodes can be merged to form a classic prestack depth migrated section. A variant of the present invention is to simulate seismic traces in the time domain. The inner core of the invention, with application of scattering wavenumbers to build filters to apply to target models, can be applied in other domains, such as Ground Penetrating Radar, and possibly acoustical and medical imaging.
|
