Applications of low-rank compressed seismic data to full-waveform inversion and extended image volumes

Abstract

Conventional oil and gas fields are increasingly difficult to explore and image, resulting in the call for more complex wave-equation-based inversion algorithms that require dense long-offset samplings. Consequently, there is an exponential growth in the size of data volumes and prohibitive demands on computational resources. We propose a method to compress and process seismic data directly in a low-rank tensor format, which drastically reduces the amount of storage required to represent the data. Seismic data exhibits low-rank structure in a particular transform domain, which can be exploited to compress the dense data in one extremely storage-efficient tensor format when the data is fully sampled, or interpolated when the data has missing entries. In either case, once our data is represented in its compressed tensor form, we propose an algorithm to extract source or receiver gathers directly from the compressed parameters. This extraction process can be done on-the-fly directly on the compressed data and does not require scanning through the entire dataset in order to form shot gathers. We apply this shot-extraction technique in the context of stochastic full-waveform inversion as well as forming full subsurface image gathers through probing techniques and demonstrate the minor differences between using the full and compressed data, while drastically reducing the total memory costs.

Seismic wavefield imaging of Earth’s interior across scales

“…As discussed previously, current FWI algorithms often invoke quasi-Newton local optimization methods, which are prone to getting 'stuck' in local minima of the misfit function. But as FWI becomes faster owing to advances in computing and more intelligent algorithms, for example, source encoding or sparsity promotion and compressive sensing [255][256][257][258][259] , the possibility of stochastic inversion using Bayesian techniques [144][145][146] becomes enticing. Global search methods based on Bayes' theorem 105 provide an entire posterior model distribution, overcoming issues associated with the single 'optimal' models currently used.…”

Seismic wavefield reconstruction using a pre-conditioned wavelet–curvelet compressive sensing approach

“…The tension between periods of high data acquisition and continually advancing computational storage and processing capabilities, has lead to cycles of interest in seismic data compression as a means to reduce the computational burden (e.g. Villasenor et al 1996;Da Silva et al 2019). The majority of these studies have focused on structured seismic data volumes, potentially with some missing elements, and so have achieved high compression ratios while maintaining fidelity-a peculiarity of the current study is that our algorithm is targeted at unstructured data volumes, necessitating the mixed-type waveletcurvelet decomposition described above.…”