Seismic Data Compression Using Deep Learning

Abstract

The exponential growth of the size of seismic data recorded in seismic surveys and real time data monitoring makes seismic data compression strongly demanded. Furthermore, compression will lead to an efficient use of the bandwidth assigned for the communication link between the seismic stations and the main center. In this paper, two convolutional autoencoders (CAEs) are proposed for seismic data compression. The two algorithms are mainly based on the convolutional neural network (CNN), which has the capability to compress the seismic data into feature representations, thereby allowing the decoder to perfectly reconstruct the input seismic data. The results show that the first model is efficient at low compression ratios (CRs), while the second model improves the signal-to-noise ratio (SNR) from about 3 dB to 12 dB compared to the other benchmark algorithms at moderate and high CRs. INDEX TERMS Convolutional autoencoders (CAE), Deep learning, Seismic data compression.

Analysis of Information Availability for Seismic and Volcanic Monitoring Systems: A Review

“…From selected articles, most of these date from 2015 to the present. The trend shows that between 2016 and 2021 seismological organizations have emphasized improving seismic data transmission by including quality parameters and an evaluation of seismic networks such as [ 26 , 31 , 36 ]. In the same way, for the areas of data acquisition and processing, as well as early warning trends, there are many studies which demonstrate a rising trend over the last 4 years, such as [ 43 , 46 , 49 ].…”

“…Regarding data compression in data transmission, Li, Huailiang et al [ 25 ] used the Lempel–Ziv–Markov chain algorithm (LZMA) and the deflate algorithm to decrease the size of SEG-Y files, and, in the experimental results, showed that the improved algorithm could provide better compression and compatibility than algorithms of standards established by the Geophysical Exploration Society (SEG) and the Steim2 compression algorithm proposed by the Standard for the Exchange of Earthquake Data (SEED). Helal et al [ 36 ] used a convolutional neural network (CNN) focused on a low-compression ratio (CRs) and a signal-to-noise ratio (SNR) to improve high and moderate seismic data CRs. Their CNN model used Keras with TensorFlow backend and Google COLAB GPU.…”

“…In this literature review, we mentioned that several studies applied new methodologies such as machine learning [ 31 , 41 ], deep learning [ 49 , 66 , 72 ], convolutional neural networks [ 36 , 61 ], among others. Mainly, they have focused on the processing and post-processing of feature data and the interpretation of seismic signals.…”

“…The compression algorithm was developed with the Lempel–Ziv–Markov chain algorithm, providing experimental results. Helal et al [ 36 ] proposed a seismic data compression model through the convolutional neural network (CNN). The main goal was to contribute to memory optimization in transmission equipment, and to preserve seismic information for rebuilding.…”

Dual‐sensor wavefield separation in a compressed domain using parabolic dictionary learning

“…Schiavon et al (2020) proposed a deep autoencoder to compress post-stack seismic data. Helal et al (2021) proposed two convolutional autoencoders, where the first model is adapted to low compression rates (CRs), whereas the second model is more efficient when the user needs to reach high CR. These methods transform the input seismic data into feature representations which are sparse enough to allow good compression performance.…”