Deep learning can automate the creation of seismic velocity models
In short: Researchers have developed a new method using a fully convolutional neural network to build seismic velocity models directly from raw seismograms. By training on large datasets, the network can automatically extract useful features and estimate models in just seconds, potentially reducing the computational time required for geophysical inversions.
Traditional techniques like full-waveform inversion are computationally expensive and rely heavily on human intervention and initial assumptions. This study used numerical experiments on synthetic models to demonstrate that a supervised deep-learning approach can establish a nonlinear projection from seismic data to velocity models.
While the training phase is time-intensive, the prediction stage is nearly instantaneous. However, the study's findings are based on synthetic models, and the researchers noted challenges such as the lack of low-frequency data in training sets.
Why it matters: AI-driven inversion could dramatically reduce the time needed for geophysical imaging once a generalized network is built.
Opens the Intellect+ feed with this card first. Swipe up for the next one.
Sources
- Summary by Intellect · source: Research: Yang and Ma (2019). Deep-learning inversion: A next-generation seismic velocity model building method. Geophysics.
To cite this card, link to https://intellectplus.co.uk/app/nature-research-deep-learning-can-build-accurate-seismic-models-faster-than.
How we source, check and correct cards: Editorial standards.
General information, not professional advice. Spotted an error? Flag the card or see corrections and complaints.
Intellect Plus is a visual educational discovery platform for exploring fascinating facts and ideas. Brain Food, Not Brain Rot. What is Intellect Plus?
Related
More in Nature · Latest · A–Z · All topics
