Large-Scale 3D Ground-Motion Synthesis with Physics-Inspired Latent Operator Flow Matching
arXiv:2603.17403v1 Announce Type: new Abstract: Earthquake hazard analysis and design of spatially distributed infrastructure, such as power grids and energy pipeline networks, require scenario-specific ground-motion time histories with realistic frequency content and spatiotemporal coherence. However, producing the large ensembles needed for uncertainty quantification with physics-based simulations is computationally intensive and impractical for engineering workflows. To address this challenge, we introduce Ground-Motion Flow (GMFlow), a physics-inspired latent operator flow matching framework that generates realistic, large-scale regional ground-motion time-histories conditioned on physical parameters. Validated on simulated earthquake scenarios in the San Francisco Bay Area, GMFlow generates spatially coherent ground motion across more than 9 million grid points in seconds, achieving a 10,000-fold speedup over the simulation workflow, which opens a path toward rapid and uncertaint
arXiv:2603.17403v1 Announce Type: new Abstract: Earthquake hazard analysis and design of spatially distributed infrastructure, such as power grids and energy pipeline networks, require scenario-specific ground-motion time histories with realistic frequency content and spatiotemporal coherence. However, producing the large ensembles needed for uncertainty quantification with physics-based simulations is computationally intensive and impractical for engineering workflows. To address this challenge, we introduce Ground-Motion Flow (GMFlow), a physics-inspired latent operator flow matching framework that generates realistic, large-scale regional ground-motion time-histories conditioned on physical parameters. Validated on simulated earthquake scenarios in the San Francisco Bay Area, GMFlow generates spatially coherent ground motion across more than 9 million grid points in seconds, achieving a 10,000-fold speedup over the simulation workflow, which opens a path toward rapid and uncertainty-aware hazard assessment for distributed infrastructure. More broadly, GMFlow advances mesh-agnostic functional generative modeling and could potentially be extended to the synthesis of large-scale spatiotemporal physical fields in diverse scientific domains.
Executive Summary
This article proposes Ground-Motion Flow (GMFlow), a physics-inspired latent operator flow matching framework that generates realistic, large-scale regional ground-motion time-histories conditioned on physical parameters. Validated on simulated earthquake scenarios in the San Francisco Bay Area, GMFlow achieves a 10,000-fold speedup over the simulation workflow, making it a promising tool for rapid and uncertainty-aware hazard assessment of distributed infrastructure. The framework's mesh-agnostic functional generative modeling capabilities also extend its potential applications to the synthesis of large-scale spatiotemporal physical fields in diverse scientific domains.
Key Points
- ▸ GMFlow is a physics-inspired latent operator flow matching framework that generates realistic ground-motion time-histories
- ▸ The framework achieves a 10,000-fold speedup over the simulation workflow
- ▸ GMFlow is validated on simulated earthquake scenarios in the San Francisco Bay Area
Merits
Speed and Efficiency
GMFlow's ability to generate large-scale ground-motion time-histories in seconds, compared to the simulation workflow's impractically long computation time, is a significant advantage in earthquake hazard analysis and design of spatially distributed infrastructure.
Mesh-Agnostic Functional Generative Modeling
GMFlow's mesh-agnostic functional generative modeling capabilities enable its potential applications to the synthesis of large-scale spatiotemporal physical fields in diverse scientific domains, making it a versatile tool.
Demerits
Limited Validation Data
The article's validation of GMFlow is limited to simulated earthquake scenarios in the San Francisco Bay Area, which may not be representative of other regions or earthquake scenarios.
Lack of Comparison to Other Methods
The article does not provide a comparison of GMFlow to other methods for generating ground-motion time-histories, making it difficult to assess its relative effectiveness and advantages.
Expert Commentary
The article proposes a promising new tool for generating realistic ground-motion time-histories, which has significant implications for earthquake hazard analysis and design of spatially distributed infrastructure. While the article's validation is limited and the comparison to other methods is lacking, the framework's mesh-agnostic functional generative modeling capabilities make it a versatile tool with potential applications in diverse scientific domains. Further research is needed to fully evaluate the effectiveness and limitations of GMFlow, as well as its potential applications and implications.
Recommendations
- ✓ Future research should focus on expanding the validation of GMFlow to other regions and earthquake scenarios, as well as comparing its performance to other methods for generating ground-motion time-histories.
- ✓ The development of GMFlow and similar tools should be supported by further research and development, as well as the establishment of standards and best practices for their use in earthquake hazard analysis and design of spatially distributed infrastructure.
