SubsurfaceGen: Procedural Generation of Field-Scale Earth Models and Seismic Data

2026-05-28

AI Summary

This study addresses the critical lack of training data in machine learning–driven full-waveform inversion (ML-FWI) that simultaneously captures spatial scale, geological diversity, and physical fidelity. To this end, we propose SubsurfaceGen—the first GPU-accelerated, procedural generation framework capable of producing field-scale, geologically diverse, and physically realistic 3D subsurface velocity models along with their corresponding seismic data. By integrating procedural modeling, wave-equation simulation, and neural operators, we construct a paired dataset comprising 4,276 2D velocity slices and associated wavefields and shot gathers across 42 real-world geological scenarios spanning six distinct geological settings. This benchmark exposes the generalization failures of current ML-FWI methods at field scale and establishes a scalable, reproducible standard for future research.

Abstract

Full waveform inversion (FWI) is the gold standard for subsurface imaging, with applications from carbon sequestration to energy and mineral exploration to earthquake hazard assessment. Machine learning approaches to FWI need field-scale, geologically diverse, and physically realistic training data, but existing resources such as Marmousi, SEAM, and OpenFWI fall short on spatial extent, temporal extent, geological diversity, and physical realism. We address these limitations with SubsurfaceGen, a GPU-accelerated generator for 3D velocity models and seismic data. Along with SubsurfaceGen, we release a paired dataset of 4,276 2D velocity slices, 5 s wavefields, and 8 s shot gathers drawn from 42 realistic, field-scale 3D velocity models, each spanning 10 km x 10 km laterally and 6.19 km deep at 10 m resolution. The dataset spans six geological settings -- four built with SubsurfaceGen and two drawn from prior sources -- relevant for carbon sequestration and hydrocarbon exploration. We use this dataset to evaluate neural operators on wavefield prediction and encoder-decoders on end-to-end velocity inversion, holding out one geological setting for out-of-distribution testing. These experiments surface failure modes at field-scale and demonstrate how SubsurfaceGen and the associated dataset can impact ML-based FWI.

Problem

• Full waveform inversion
• Training data
• Geological diversity
• Physical realism
• Field-scale

Innovation

• procedural generation
• full waveform inversion
• field-scale modeling
• GPU-accelerated simulation
• geologically diverse dataset