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/

Zendy Plus

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

Zendy Tools

Zendy Open

Achieving desirable receiver sampling in ocean bottom acquisition is oftennot possible because of cost considerations. Assuming adequate source samplingis available, which is achievable by virtue of reciprocity and the use ofmodern randomized (simultaneous-source) marine acquisition technology, we arein a position to train convolutional neural networks (CNNs) to bring thereceiver sampling to the same spatial grid as the dense source sampling. Toaccomplish this task, we form training pairs consisting of densely sampled dataand artificially subsampled data using a reciprocity argument and theassumption that the source-site sampling is dense. While this approach hassuccessfully been used on the recovery monochromatic frequency slices, itsapplication in practice calls for wavefield reconstruction of time-domain data.Despite having the option to parallelize, the overall costs of this approachcan become prohibitive if we decide to carry out the training and recoveryindependently for each frequency. Because different frequency slices shareinformation, we propose the use the method of transfer training to make ourapproach computationally more efficient by warm starting the training with CNNweights obtained from a neighboring frequency slices. If the two neighboringfrequency slices share information, we would expect the training to improve andconverge faster. Our aim is to prove this principle by carrying a series ofcarefully selected experiments on a relatively large-scale five-dimensionaldata synthetic data volume associated with wide-azimuth 3D ocean bottom nodeacquisition. From these experiments, we observe that by transfer training weare able t significantly speedup in the training, specially at relativelyhigher frequencies where consecutive frequency slices are more correlated.

Transfer learning in large-scale ocean bottom seismic wavefield reconstruction