This section focuses on transforming raw seismic data into high-fidelity subsurface images through physics-driven and data-driven workflows. Topics include preprocessing (deblending, denoising, demultiple, amplitude correction), velocity model building (tomography, FWI), imaging algorithms (Kirchhoff, RTM, LSM, LSRTM), and resolution enhancement (deconvolution, Q-compensation). Emphasis is placed on multi-dimensional processing (3D/4D, 5D interpolation), machine learning integration (for signal separation, velocity prediction, or automated QC), elastic/viscoacoustic imaging, and uncertainty quantification. Submissions must demonstrate technical novelty and practical impact on interpretability, artifact suppression, or structural/stratigraphic resolution. Comparative analyses of methods are welcome.
Microseismic event locations using grid-searching method and Newton–Raphson-based optimizer
A physics-constrained sparse basis learning method for mixed noise suppression
GMLAN: Grouped-residual and multi-scale large-kernel attention network for seismic image super-resolution
Fully automatic differentiation with coupling deep neural networks for full-waveform inversion
Dual-branch dense network for seismic background noise elimination
Microseismic event picking and classification for hot dry rock hydraulic fracturing monitoring using SeisFormer
U-STDRNet: A unified model integrating swin transformer and residual dense network for seismic image super-resolution and denoising
Unscaled generalized S-transform and its applications on seismic attenuation delineation: A case study in the Ordos Basin, Northwest China