English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Noise in seismic data arises from numerous sources and is continuallyevolving. The use of supervised deep learning procedures for denoising ofseismic datasets often results in poor performance: this is due to the lack ofnoise-free field data to act as training targets and the large difference incharacteristics between synthetic and field datasets. Self-supervised,blind-spot networks typically overcome these limitation by training directly onthe raw, noisy data. However, such networks often rely on a random noiseassumption, and their denoising capabilities quickly decrease in the presenceof even minimally-correlated noise. Extending from blind-spots to blind-maskscan efficiently suppress coherent noise along a specific direction, but itcannot adapt to the ever-changing properties of noise. To preempt the network'sability to predict the signal and reduce its opportunity to learn the noiseproperties, we propose an initial, supervised training of the network on afrugally-generated synthetic dataset prior to fine-tuning in a self-supervisedmanner on the field dataset of interest. Considering the change in peaksignal-to-noise ratio, as well as the volume of noise reduced and signalleakage observed, we illustrate the clear benefit in initialising theself-supervised network with the weights from a supervised base-training. Thisis further supported by a test on a field dataset where the fine-tuned networkstrikes the best balance between signal preservation and noise reduction.Finally, the use of the unrealistic, frugally-generated synthetic dataset forthe supervised base-training includes a number of benefits: minimal priorgeological knowledge is required, substantially reduced computational cost forthe dataset generation, and a reduced requirement of re-training the networkshould recording conditions change, to name a few.

Transfer learning for self-supervised, blind-spot seismic denoising