Training a generalizable diffusion model for seismic data processing using a large-scale open-source waveform dataset

TL;DR

This paper presents SWAN, a large-scale seismic waveform dataset, and demonstrates that diffusion models trained on it achieve state-of-the-art results in seismic data reconstruction, outperforming existing methods.

Contribution

Introduction of SWAN, a comprehensive seismic waveform dataset, and development of a diffusion model that achieves superior generalization and performance in seismic data processing tasks.

Findings

Diffusion models trained on SWAN outperform baselines in seismic reconstruction.

SWAN provides a unified benchmark for diverse seismic data scenarios.

Diffusion architectures show strong potential for robust seismic processing.

Abstract

We introduce the Seismic Waveforms dataset for Automatic Neural-network processing (SWAN), a comprehensive and standardized benchmark designed to advance data-driven seismic signal processing. SWAN aggregates diverse synthetic and real seismic waveforms spanning a wide range of geological structures, noise conditions, propagation environments, and acquisition geometries, providing a unified foundation for training highly generalizable models. Leveraging this dataset, we develop and evaluate a conditionally constrained residual diffusion model for core seismic processing tasks, focusing on missing-trace reconstruction. Extensive experiments demonstrate that diffusion models trained on SWAN achieve state-of-the-art performance across heterogeneous testing scenarios, outperforming leading deep-learning and physics-based baselines on both synthetic benchmarks and field data examples. The results highlight SWAN's value as both a scalable training corpus and a rigorous evaluation framework, and illustrate the strong potential of diffusion-based architectures for robust, generalizable seismic data processing.