Fault-conditioned Seismic Image Generation using Denoising Diffusion Probabilistic Modeling and Neural Style Transfer
Abstract
Seismic data augmentation is critical for training robust deep learning (DL) models for fault detection and characterization, yet conventional approaches remain inadequate. Geometric transforms, e.g., flips or rotations, and noise injection fail to capture realistic seismic textures, while physics-based forward modeling is computationally expensive and often poorly aligned with the statistical characteristics of field data. We present a denoising diffusion probabilistic model (DDPM) framework that learns the high-dimensional distribution of seismic images directly from field data via iterative noise injection and learned denoising. We develop two variants: an unconditional model (U-DDPM) that generates realistic post-stack seismic patches and a fault-conditioned model (FC-DDPM) that produces diverse seismic realizations that honor user-specified fault geometries encoded as binary masks. Three open-source field datasets with differing amplitude scales and processing histories are jointly standardized for training, demonstrating cross-survey generalization. Generated images from U-DDPM reproduce the dominant textural patterns, amplitude distributions, and reflector continuity observed in real seismic data. FC-DDPM maintains structural fidelity to imposed fault masks while introducing realistic variability in surrounding seismic textures, enabling one-to-many realizations from a single structural template. Frequency-wavenumber spectral analysis confirms that synthetic samples preserve the spatiotemporal bandwidth and dip content of the training distribution. The proposed framework provides a practical, controllable augmentation mechanism for expanding limited fault-labeled training sets without computationally expensive forward modeling or survey-specific parameter tuning, directly addressing a key bottleneck in DL-based seismic interpretation.
Related
- Scenario-conditioned flow matching for probabilistic generation of three-component ground-motion waveforms
- Broadband Ground-Motion Synthesis by Conditioning Denoising Diffusion Models on Low-Frequency Waveforms
- Preserving Probabilistic Meaning in Weather-to-Text Generation
- Bayesian Joint Velocity and Impedance Inversion via Diffusion Models Conditioned on Common Image Gathers
- Diffusion-Based Refinement for Kilometer-Scale Probabilistic Precipitation Nowcasting
BibTeX
@misc{agbaje2026faultconditioned,
title = {{Fault-conditioned Seismic Image Generation using Denoising Diffusion Probabilistic Modeling and Neural Style Transfer}},
author = {Tolulope Agbaje and Nam Pham and Altay Sansal and Yangkang Chen},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.10788},
eprint = {2610.10788},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.10788},
}
RIS
TY - UNPB
TI - Fault-conditioned Seismic Image Generation using Denoising Diffusion Probabilistic Modeling and Neural Style Transfer
AU - Tolulope Agbaje
AU - Nam Pham
AU - Altay Sansal
AU - Yangkang Chen
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.10788
UR - https://arxiv.org/abs/2610.10788
AB - Seismic data augmentation is critical for training robust deep learning (DL) models for fault detection and characterization, yet conventional approaches remain inadequate. Geometric transforms, e.g., flips or rotations, and noise injection fail to capture realistic seismic textures, while physics-based forward modeling is computationally expensive and often poorly aligned with the statistical characteristics of field data. We present a denoising diffusion probabilistic model (DDPM) framework that learns the high-dimensional distribution of seismic images directly from field data via iterative noise injection and learned denoising. We develop two variants: an unconditional model (U-DDPM) that generates realistic post-stack seismic patches and a fault-conditioned model (FC-DDPM) that produces diverse seismic realizations that honor user-specified fault geometries encoded as binary masks. Three open-source field datasets with differing amplitude scales and processing histories are jointly standardized for training, demonstrating cross-survey generalization. Generated images from U-DDPM reproduce the dominant textural patterns, amplitude distributions, and reflector continuity observed in real seismic data. FC-DDPM maintains structural fidelity to imposed fault masks while introducing realistic variability in surrounding seismic textures, enabling one-to-many realizations from a single structural template. Frequency-wavenumber spectral analysis confirms that synthetic samples preserve the spatiotemporal bandwidth and dip content of the training distribution. The proposed framework provides a practical, controllable augmentation mechanism for expanding limited fault-labeled training sets without computationally expensive forward modeling or survey-specific parameter tuning, directly addressing a key bottleneck in DL-based seismic interpretation.
ER -