Reproduce “$\mathbf{\mathbb{E}^{FWI}}$: Multi-parameter Benchmark Datasets for Elastic Full Waveform Inversion of Geophysical Properties” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Shihang Feng, Hanchen Wang, Chengyuan Deng, Yinan Feng, Yanhua Liu, Min Zhu, Peng Jin, Yinpeng Chen and 1 more

Los Alamos National Laboratory and collaborators introduce EFWI, an open-source, multiparameter benchmark dataset for elastic Full Waveform Inversion, featuring 0.69 TB across 168,000 samples of P-wave and S-wave velocity maps and multi-component seismic data. This resource fills a critical gap for training and evaluating data-driven methods in subsurface characterization, particularly for carbon-zero and new energy exploration.

Key insights

  • The intrinsic coupling between P-wave and S-wave velocities is crucial for accurate elastic FWI, as ignoring this relationship leads to substantial degradation in inversion performance and misplacement of subsurface features.
  • Performance of data-driven FWI models consistently declines with increasing geological complexity in the subsurface models, highlighting challenges in generalization.
  • While P-wave and S-wave velocities can be directly predicted, indirectly calculated parameters like Poisson's ratio often show lower structural similarity, indicating sensitivity to combined errors in primary predictions.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.