No abstract

Time shift estimation is a key issue in many areas of seismic exploration such as time-lapse studies and traveltime tomography. Automated estimation is useful because it enables applications such as time strain analysis and automated velocity model building. A commonly used automated estimation algorithm is the windowed crosscorrelation. Unfortunately, this method can be inaccurate in areas where time shifts vary significantly over short intervals. Algorithms based on mismatch minimization have been proposed to remedy this problem. All the mentioned methods suffer from cycle skipping in the face of large time shifts. I, therefore, used the dynamic time warping (DTW) algorithm as an alternative method to estimate time shifts. I claimed the time shift estimates produced by DTW are of higher quality than those produced by the previous methods, especially with regard to the cycle-skipping problem. I found that DTW always is able to perfectly restore time shifts applied to traces when the changes are purely temporal. I investigated its accuracy when amplitude changes are present, and found that the algorithm is well suited for time-lapse analysis. Finally, we determined that DTW is resilient against cycle-skipping errors, in the presence of large time shifts and with significant changes in the data, and thus a candidate for use in automated velocity model-building methods, such as waveform tomography or full-waveform inversion.

The difference in computational power between the few- and multicore architectures represented by central processing units (CPUs) and graphics processing units (GPUs) is significant today, and this difference is likely to increase in the years ahead. GPUs are, therefore, ever more popular for applications in computational physics, such as wave modeling. Finite-difference methods are popular for wave modeling and are well suited for the GPU architecture, but developing an efficient and capable GPU implementation is hindered by the limited size of the GPU memory. I revealed how the out-of-core technique can be used to circumvent the memory limit on the GPU, increasing the available memory to that of the CPU (the main memory) instead, with no significant computational overhead. This approach has several advantages over a parallel scheme in terms of applicability, flexibility, and hardware requirements. Choices in the numerical scheme — the numerical differentiators in particular — also greatly affect computational efficiency. These factors are considered explicitly for GPU implementations of wave modeling because GPUs are special purpose with a visible architecture.