Яндекс.Метрика

Kirill Gadylshin, Vadim Lisitsa, Kseniia Gadylshina, Dmitry Vishnevsky

: Computational Science and Its Applications - ICCSA 2023 Workshops. 23rd International Conference (Athens, Greece, July 3-6, 2023)
: Computational Science and Its Applications - ICCSA 2023 Workshops. 23rd International Conference (Athens, Greece, July 3-6, 2023)

This study proposes using an NDM-net in the frequency domain to mitigate numerical dispersion in seismic data, a major problem in the petroleum industry. Frequency-domain processing has several advantages over time-domain processing, including more efficient extraction of information from large datasets, noise removal, and correction for amplitude and phase distortions. The proposed approach can significantly improve the exploration and characterization of hydrocarbon reservoirs, with Full Waveform Inversion (FWI) being a prominent example. To evaluate the proposed method, we applied it to a realistic synthetic BP numerical example, resulting in an average reduction of numerical dispersion by a factor of three and an order-of-magnitude increase in calculation performance. The proposed approach enables us to process the entire seismogram simultaneously, in contrast to the time-domain NDM-net implementation and thus can be applied to 3D seismic modeling. Overall, the use of NDM-net in the frequency domain can significantly improve the quality and efficiency of seismic data processing, enhancing our understanding of hydrocarbon reservoirs and facilitating more accurate exploration and characterization of subsurface structures.
индекс в базе ИАЦ: 027939