1 June 2014 in Computers & Geosciences
1 May 2021 in Journal of Petroleum Science and Engineering
19 August 2014 in Geophysics
To obtain subsurface information from onshore seismic exploration data using full-waveform inversion (FWI) based on the acoustic wave equation, elastic waves, such as ground roll and mode-converted waves, should be suppressed through heavy preprocessing. However, such preprocessing deforms not only the elastic waves but also the acoustic waves. Moreover, it is not easy to separate body waves from surface waves in seismic traces. For these reasons, we need to generate both types of waves in the modeling step to obtain seismic waves that are similar to real seismic waves. Therefore, elastic FWI using the elastic wave equation is necessary to achieve a more accurate FWI. In addition, elastic FWI can provide better geologic information than acoustic FWI because it inverts the P-wave velocity, S-wave velocity, and density. Laplace-Fourier-domain elastic FWI is an effective method because it inverts these multiple parameters and can be applied to real seismic data that lack low-frequency components. However, the conventional Laplace-Fourier-domain FWI requires a matrix solver with a huge memory cost to perform the modeling in the Laplace-Fourier-domain, and memory usage becomes more intensive in the elastic case. In the present study, we combined time-domain wave propagation modeling and Laplace-Fourier-domain elastic FWI to invert multiple parameters with less memory cost. By using time-domain modeling, which does not require a matrix solver, we obtained the forward and adjoint wavefields with less memory cost. The residuals between the recorded and modeled data, the virtual sources, the Hessian matrices, and the gradient directions were calculated in the Laplace-Fourier domain. To validate the proposed algorithm, we performed numerical tests with Model 94 synthetic data and Benjamin Creek real seismic data.
11 November 2020 in Ocean Science
Seismic oceanography (SO) acquires water column reflections using controlled source seismology and provides high lateral resolution that enables the tracking of the thermohaline structure of the oceans. Most SO studies obtain data using air guns, which can produce acoustic energy below 100 Hz bandwidth, with vertical resolution of approximately 10 m or more. For higher-frequency bands, with vertical resolution ranging from several centimeters to several meters, a smaller, low-cost seismic exploration system may be used, such as a sparker source with central frequencies of 250 Hz or higher. However, the sparker source has a relatively low energy compared to air guns and consequently produces data with a lower signal-to-noise (S∕N) ratio. To attenuate the random noise and extract reliable signal from the low S∕N ratio of sparker SO data without distorting the true shape and amplitude of water column reflections, we applied machine learning. Specifically, we used a denoising convolutional neural network (DnCNN) that efficiently suppresses random noise in a natural image. One of the most important factors of machine learning is the generation of an appropriate training dataset. We generated two different training datasets using synthetic and field data. Models trained with the different training datasets were applied to the test data, and the denoised results were quantitatively compared. To demonstrate the technique, the trained models were applied to an SO sparker seismic dataset acquired in the Ulleung Basin, East Sea (Sea of Japan), and the denoised seismic sections were evaluated. The results show that machine learning can successfully attenuate the random noise in sparker water column seismic reflection data.
1 December 2015 in Journal of Applied Geophysics
1 August 2019 in Journal of Asian Earth Sciences