Large-Scale Computation for Seismic Applications

本研究计划的重点是发展解决声波和弹性波动方程的改进技术。该工作可应用于地震勘探、无损检测、声纳解释等领域。当前的目标是改进所谓正问题的数值方法，在正问题中，假设初始边界数据和材料属性是已知的。然而，该研究的最终目标是以迭代的方式使用精确的正演求解器来解决逆问题，在逆问题中，根据观察到的介质对地震波的响应，寻求材料特性（波速）的估计。该项目代表了怀俄明州早期研究的延续，其中开发了用于在三维空间中求解声学和弹性波动方程的代码。该项目涉及几种数值方法。其中包括具有不同波速的地层之间界面的表示，将在高度非均质介质中产生精确离散化的空间网格的构建，有效的时间步进方案和相关的数值线性代数的设计，以及数值程序对并行和矢量结构机器的适应性。该项目的另一个重要元素是继续开发面向对象的图形软件，以帮助大规模模型结果的可视化。

This research project focuses on the development of improved techniques for solving the acoustic and elastic wave equations. The work has applications to seismic exploration, nondestructive testing, sonar interpretation, and other fields. The immediate aim is to enhance numerical methods for the so-called forward problem, in which initial-boundary data and material properties are assumed known. However, the research has the ultimate objective of using accurate forward solvers in an iterative fashion to attack the inverse problem, in which one seeks estimates of material properties (wave velocities) given observed responses of the medium to seismic waves. The project represents a continuation of earlier research at Wyoming, in which codes were developed for solving the acoustic and elastic wave equations in three space dimensions. The project addresses several numerical approaches. Among these are the representation of interfaces between strata having different wave velocities, the construction of spatial grids that will yield accurate discretizations in highly heterogeneous media, the design of efficient time-stepping schemes and associated numerical linear algebra, and the adaptation of numerical procedures to parallel- and vector architecture machines. Another important element of the project is the continued development of object-oriented graphics software to aid in the visualization of large-scale model results.