Slob Evert, Wapenaar Kees, Treitel Sven
Department of Geoscience and Engineering Delft University of Technology P.O. Box 5048 Delft GA 2600 The Netherlands.
Tridekon Oklahoma USA.
Geophys Prospect. 2020 Jun;68(5):1425-1442. doi: 10.1111/1365-2478.12946. Epub 2020 Mar 17.
Acoustic inversion in one-dimension gives impedance as a function of travel time. Inverting the reflection response is a linear problem. Recursive methods, from top to bottom or vice versa, are known and use a fundamental wave field that is computed from the reflection response. An integral over the solution to the Marchenko equation, on the other hand, retrieves the impedance at any vertical travel time instant. It is a non-recursive method, but requires the zero-frequency value of the reflection response. These methods use the same fundamental wave field in different ways. Combining the two methods leads to a non-recursive scheme that works with finite-frequency bandwidth. This can be used for target-oriented inversion. When a reflection response is available along a line over a horizontally layered medium, the thickness and wave velocity of any layer can be obtained together with the velocity of an adjacent layer and the density ratio of the two layers. Statistical analysis over 1000 noise realizations shows that the forward recursive method and the Marchenko-type method perform well on computed noisy data.
一维声学反演给出了作为走时函数的阻抗。对反射响应进行反演是一个线性问题。递归方法,从上到下或反之亦然,是已知的,并且使用从反射响应计算得到的基本波场。另一方面,对马尔琴科方程的解进行积分,可以在任何垂直走时瞬间获取阻抗。这是一种非递归方法,但需要反射响应的零频率值。这些方法以不同方式使用相同的基本波场。将这两种方法结合会产生一种适用于有限频率带宽的非递归方案。这可用于目标导向反演。当沿着水平分层介质上的一条线有反射响应可用时,可以一起获得任何层的厚度和波速,以及相邻层的速度和两层的密度比。对1000次噪声实现的统计分析表明,前向递归方法和马尔琴科型方法在计算得到的噪声数据上表现良好。